Cover page

Pillars of Creation

(Image source : https://www.planetary.org/space-images/jwst-pillars-of-creation )

This is the image of the Eagle Nebula at a distance of 6500 light years as seen through the James Web Space

Telescope. This star-forming region was made famous when the Hubble Space Telescope imaged it in 1995. It

is where new stars are born! The dust cloud that looks like a monster consists of microscopic dust particles and

hydrogen gas which form the raw materials for the new born stars! The shining head are composed of newborn

young stars. Image courtesy: NASA, ESA, CSA, STScI; Joseph DePasquale (STScI), Anton M. Koekemoer

(STScI), Alyssa Pagan (STScI)

Managing Editor Chief Editor Editorial Board Correspondence

Ninan Sajeeth Philip Abraham Mulamootil K Babu Joseph The Chief Editor

Ajit K Kembhavi airis4D

Geetha Paul Thelliyoor - 689544

Arun Kumar Aniyan India

Jorunal Publisher Details

Publisher : airis4D, Thelliyoor 689544, India

Website : www.airis4d.com

Email : nsp@airis4d.com

Phone : +919497552476

Editorial

by Fr Dr Abraham Mulamoottil

airis4D, Vol.1, No.5, 2023

www.airis4d.com

This edition of airis4D discusses various topics, including Bayesian inference, Natural Language Processing

(NLP), Solar Flares, Cataclysmic Variable Stars, the Barcode of Life Project, and The Ten Years of Science.

Bayesian inference is an essential statistical inference technique that updates the probability of a hypothesis

as new data becomes available. The technique is useful in machine learning for making predictions based

on prior beliefs, and the posterior distribution of predictors can be updated based on new evidence. Blesson

George rightly points out that Bayesian inference is an important technique in statistics, useful in various ﬁelds,

including medicine, business, economics, and social sciences.

Jinsu Ann Mathew’s article of Natural Language Processing (NLP) provides a good overview of the

ﬁeld, including its three main technologies, NLU, NLG, and NLP. NLU enables machines to comprehend and

interpret human language, while NLG uses the information gathered from NLU to generate a natural-sounding

and situation-appropriate response. The article notes that without NLU, machines would struggle to process

human language, limiting their ability to extract insights and value from this vast source of information. The

article’s discussion of NLP is brief, but it is clear that it is an essential technology that enables machines to

understand and generate human language.

The article on solar ﬂares provides an interesting historical perspective on the topic, tracing the evolution

of scientiﬁc understanding of the Sun from its worship as a deity to a scientiﬁc object of study. Linn Abraham

notes that solar ﬂares emit intense bursts of electromagnetic radiation that can cause broadcast interference and

power outages on Earth, making it important to understand them to protect power grids, satellites, and astronauts

during spacewalks. While there is no immediate need to predict solar ﬂares, it is crucial to understand them to

develop strategies for mitigating their eﬀects.

The article of Sindhu G on cataclysmic variable stars provides an informative overview of these fascinating

objects, which are binary star systems that undergo sudden and dramatic changes in their properties. The article

notes that they are actively researched by astronomers to better understand their properties and evolution. While

the article is brief, it provides an excellent introduction to the topic and is likely to spark the reader’s interest in

learning more about these objects.

Geetha Paul’s article on the Barcode of Life project highlights the project’s ambitious goal of creating a

comprehensive, digital library of DNA barcodes for every species on Earth. The project’s potential applications

are vast, including identifying species in ecology, agriculture, medicine, and food safety, among others. The

article notes that the project’s main challenge is the sheer scale of the project, with an estimated 8.7 million

species on Earth. However, the Barcode of Life project has already made signiﬁcant progress in documenting

and preserving the world’s biodiversity.

Robin Jacob Roy calls to understand and prepare for the dangers posed by tropical cyclones. Tropical

cyclones are fueled by heat and moisture from warm ocean surfaces, and characterized by a low-pressure center,

strong winds, and heavy rain. Favorable environmental factors are needed for a tropical cyclone to develop,

including warm ocean waters, unstable atmosphere, moist air, distance from the equator, and minimal wind

shear. A tropical cyclone’s structure includes the eye, eyewall, spiral rainbands, upper-level outﬂow, and sea

surface. Tropical cyclones have various hazards, including storm surges, ﬂooding, powerful winds, tornadoes,

and lightning, and preparation for these dangers is crucial. Once a tropical cyclone reaches sustained winds of at

least 74 mph, it is classiﬁed as a hurricane or typhoon, depending on the region. When these hazards combine

and interact with one another, the potential for loss of life and damage to property increases substantially.

The article ”Bringing Industrial Practices to Academia” by Arun Aniyan talks about how the work culture in

academia diﬀers from that of the industry and how bringing the discipline and work methodology of the industry

can improve the work style and throughput of academic work. The article introduces the Agile methodology

and explains its formal framework for product development. The Agile methodology emphasizes deﬁning the

problem, scope, and requirements of a product in a strict and disciplined manner, which can help in achieving

the desired outcome in academia as well. It also emphasizes dividing the solution into independent parts to set

speciﬁc milestones for execution.

”The Ten Years of Science” by Ninan Sajeeth Philip discusses signiﬁcant scientiﬁc developments across

various ﬁelds in the past ten years, including astronomy, genetics and biotechnology, microbiome research, and

neuroscience. The past decade has witnessed remarkable progress in these ﬁelds, leading to new insights into

the nature of the universe and transforming our understanding of the world.

Overall, the articles provide various topics to spark the reader’s interest in learning more about them. The

writing is clear and concise, making it accessible to a broad audience.

iii

Contents

Editorial ii

I Artiﬁcial Intelligence and Machine Learning 1

1 Bayesian Interference in Machine Learning 2

1.1 Introduction to Bayesian Inference . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2

1.2 Bayesian Inference in Machine Learning . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3

1.3 Advantages and Limitations of Bayesian Inference in Machine Learning . . . . . . . . . . . . 5

2 Unpacking NLP: Understanding the Roles of NLU and NLG 6

2.1 How NLU and NLG ﬁt into NLP . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6

2.2 Natural Language Understanding . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6

2.3 Natural Language Generation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8

II Astronomy and Astrophysics 11

1 Exploring the Explosive Side of the Sun 12

1.1 The Sun and Life on Earth . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12

1.2 Historical Record of Solar Explosions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12

1.3 Do We Need to Predict These Outbursts? . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14

1.4 How the Modern World is Becoming More Vulnerable to Solar Flares? . . . . . . . . . . . . . 14

1.5 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15

2 Cataclysmic Variable Stars 16

2.1 What are Cataclysmic Variable stars . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16

2.2 Discovery . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17

2.3 Some types of Cataclysmic Variables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18

III Biosciences 23

1 Barcode of Life: A Global Biodiversity Challenge 24

1.1 NEXT GENERATION SEQUENCING (NGS) . . . . . . . . . . . . . . . . . . . . . . . . . 27

1.2 DNA barcoding as a new tool for food traceability . . . . . . . . . . . . . . . . . . . . . . . . 31

1.3 Highlights . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31

1.4 Strengths, Weaknesses, Opportunities, and Threats for DNA barcoding, resulting from the

SWOT analysis. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31

IV Climate 33

1 Tropical Cyclones 34

CONTENTS

1.1 Formation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34

1.2 Structure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35

1.3 Classiﬁcation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36

1.4 Naming . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37

1.5 Forecasting . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37

V General 39

1 Bringing Industrial Practices to Academia 40

1.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40

1.2 Agile Research . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41

1.3 Process of Agile Execution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42

1.4 Reward function . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44

1.5 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44

2 The Ten Years of Science 46

2.1 Development across sciences . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46

2.2 Astronomy Research . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46

2.3 Gravitational Wave Detection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47

2.4 Exoplanets . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49

Part I

Artiﬁcial Intelligence and Machine Learning

Bayesian Interference in Machine Learning

by Blesson George

airis4D, Vol.1, No.5, 2023

www.airis4d.com

1.1 Introduction to Bayesian Inference

Statistical inference is the process of using data analysis techniques, statistical methods, and probability

theory to draw conclusions or inferences about a population based on a data sample. In many instances, it is

impractical or impossible to acquire data from an entire population; as a result, only a representative sample

is collected. Statistical inference enables us to draw conclusions about the larger population based on this

sample data. Often, inference permits the derivation of the properties of an underlying probability distribution.

Numerous disciplines, including medicine, business, economics, social sciences, etc., utilise statistical inference

frequently. The two main categories of statistical inference are estimation and hypothesis testing. In estimation,

sample statistics are used to estimate population parameters, and in hypothesis testing, sample data are used to

determine the probability that a hypothesis is true.

Bayesian inference is a method of statistical inference in which Bayes’ theorem is applied to revise the

probability of a hypothesis as additional evidence or data becomes available. It is based on Bayes’ theorem,

which describes the relationship between the prior probability of a hypothesis and the likelihood of the observed

data given the hypothesis.

Thomas Bayes, an English statistician of the 18th century, formulated the Bayes theorem, which is given

in its most popular form as

P (A|B) =

P (B|A)P (A)

P (B)

where

P (A|B) is the conditional probability of event A occurring given that B is true. It is deﬁned as the

’Posterior’ probability. The term ”posterior” is used to distinguish the updated probability from the

prior probability, which is the probability assigned to the parameter before observing any new data. The

posterior probability is a fundamental concept in Bayesian inference, which is a statistical framework that

allows us to update our beliefs or knowledge about a phenomenon based on new evidence or data.

P (B|A) is the conditional probability of event B occurring given that A is true. It is known as the

likelihood, which is the probability of observing a certain set of data given a speciﬁc parameter value. It

represents the degree to which the data support or favour the hypothesis and measures the compatibility

between the observed data and the hypothesis.

P (A) and P (B) are the probabilities of A and B occurring independently of one another (the marginal

probability). P (A) is known as the prior. The prior is the initial or prior probability assigned to a

1.2 Bayesian Inference in Machine Learning

parameter before observing any new data. It represents our prior belief or degree of uncertainty about the

parameter based on our background knowledge, assumptions, or previous studies.

Bayes’ theorem for hypothesis testing is typically expressed as follows:

P (H|E) =

P (E|H)P (H)

P (E)

where P (H|E) gives the probability that a hypothesis is true given the evidence, P (H) is the probability for

hypothesis being true, P (E|H ) deﬁnes the probability of seeing the evidence if the hypothesis is true and P (E)

gives the probability of observing the evidence.

1.1.1 Bayesian inference and frequentist inference

Bayesian inference and frequentist inference are two distinct methods of statistical inference, the process

of deriving conclusions from data.

Bayesian inference, which is based on Bayes’ theorem, entails revising prior beliefs about a hypothesis

or parameter based on new data to calculate a posterior probability. The posterior distribution represents

an updated hypothesis or parameter belief. Bayesian inference permits us to incorporate prior knowledge,

uncertainty, and hypotheses into our analysis, and to make probabilistic predictions and decisions based on the

available evidence.

On the other hand, frequentist inference relies on the frequentist interpretation of probability, which deﬁnes

probability as the long-run relative frequency of an event. Frequentist inference entails estimating the probability

of an event or parameter based on observed data, presuming that the observed data are a random sample from a

larger population. Frequentist inference does not rely on prior distributions or hypotheses and typically involves

evaluating hypotheses and calculating conﬁdence intervals.

The treatment of prior knowledge and uncertainty is a crucial distinction between Bayesian and frequentist

inferences. In contrast to frequentist inference, Bayesian inference explicitly incorporates prior knowledge and

uncertainty into the analysis. This can result in diﬀering conclusions or decisions based on the same data,

depending on the analyst’s preconceived notions and assumptions.

1.2 Bayesian Inference in Machine Learning

Bayesian inference can be utilised for a variety of tasks in machine learning, including parameter estimation,

model selection, regression, classiﬁcation, and clustering.

Bayesian inference can also be used for model selection and comparison, where diﬀerent models or

hypotheses are compared based on their posterior probabilities. This allows for a principled way of selecting

the best model or hypothesis given the available data and prior knowledge.

One advantage of Bayesian inference in machine learning is its ability to incorporate prior knowledge or

assumptions into the analysis, which can improve the accuracy and robustness of the results. Bayesian inference

can also provide a natural way of handling uncertainty and making probabilistic predictions or decisions.

1.2.1 Some Bayesian inference models for machine learning

1.2.1.1 Naive Bayes

Naive Bayes is a simple but eﬀective probabilistic algorithm used in machine learning for classiﬁcation

tasks. It is based on Bayes’ theorem, which states that the probability of a hypothesis (or class) given some

observed data can be computed using the prior probability of the hypothesis and the likelihood of the data

1.3 Advantages and Limitations of Bayesian Inference in Machine Learning

given the hypothesis. Naive Bayes assumes that the features (or variables) are conditionally independent given

the class, which means that the presence or absence of one feature does not aﬀect the likelihood of another

feature. Despite this strong independence assumption, Naive Bayes can perform well in practice and is often

used as a baseline model for text classiﬁcation, spam ﬁltering, and other applications where the feature space is

high-dimensional and sparse.

1.2.1.2 Linear Discriminant Analysis

Linear Discriminant Analysis (LDA) is a classic technique used in supervised learning for dimensionality

reduction and classiﬁcation tasks. Bayesian inference can be used to estimate the parameters of LDA and make

predictions based on new data.

In LDA, the goal is to ﬁnd a linear combination of the input features that maximally separates the

classes, while also minimizing the within-class variance. The resulting linear discriminants can be used for

dimensionality reduction and classiﬁcation of new data. Bayesian inference can be used to estimate the mean

and covariance matrix of the input features for each class, as well as the prior probabilities of each class. This

allows for a principled way of handling uncertainty and incorporating prior knowledge into the analysis.

1.2.1.3 Linear regression

In Bayesian linear regression, for instance, the objective is to estimate the coeﬃcients of a linear model

using training data and a prior distribution on the coeﬃcients. From a Bayesian perspective, linear regression

is formulated using probability distributions rather than point estimates. The response is assumed to be drawn

from a probability distribution rather than a distinct value.

The objective of Bayesian Linear Regression is not to determine the singular ”best” value of the model

parameters, but rather the posterior distribution of the model parameters. Not only is the response derived from

a probability distribution, but it is also presumed that the model parameters are drawn from a distribution. The

posterior probability of the model parameters is dependent on the inputs and outputs of training. Using Bayes’

theorem, the posterior distribution of the coeﬃcients can be computed by multiplying the prior distribution

by the likelihood of the data. The posterior distribution can then be utilised to make predictions or quantify

uncertainty.

1.2.1.4 Markov Chain Monte Carlo (MCMC) methods

Markov Chain Monte Carlo (MCMC) methods are a class of algorithms used in Bayesian inference to

simulate and approximate the posterior distribution of the model parameters. MCMC methods use a Markov

chain to generate a sequence of samples from the posterior distribution, where each sample is dependent on the

previous sample and is drawn from a proposal distribution. Bayesian inference in MCMC methods involves

using the simulated samples to estimate the posterior distribution of the parameters of interest.

Bayesian inference in MCMC methods can be used for a variety of applications, including parameter

estimation, model selection, and uncertainty quantiﬁcation. MCMC methods are particularly useful when the

posterior distribution is complex or high-dimensional and cannot be computed analytically. MCMC methods

can also handle non-linear and non-Gaussian models, as well as incorporate prior knowledge or constraints on

the model parameters.

1.3 Advantages and Limitations of Bayesian Inference in Machine Learning

The advantages of Bayesian inference methodologies are: Bayesian Inference oﬀers a framework for

incorporating prior information and assumptions into a model. This can enhance the model’s accuracy and

reduce the danger of overﬁtting.In addition, Bayesian Inference provides a method for quantifying uncertainty

in model predictions, which is useful for risk assessment and decision making. Bayesian Inference is also a

ﬂexible machine learning tool because it can be applied to a wide variety of models and data types.

Computational Complexity is one of the limitations of Bayesian Inference in Machine Learning. Bayesian

Inference can be computationally costly, particularly for complex models and large data sets. Prior Selection -

The choice of prior distribution can have a substantial impact on the results of Bayesian Inference, and selecting

an appropriate prior can be diﬃcult.

References

Bayes Theorem in Machine learning

Introduction to Bayesian Logistic Regression

Bayesian Inference: The Best 5 Models and 10 Best Practices for Machine Learning

Bayes theorem

Bayesian Analysis: Advantages and Disadvantages

About the Author

Blesson George is currently working as Assistant Professor of Physics at CMS College Kottayam,

Kerala. His research interests include developing machine learning algorithms and application of machine

learning techniques in protein studies.

Unpacking NLP: Understanding the Roles of

NLU and NLG

by Jinsu Ann Mathew

airis4D, Vol.1, No.5, 2023

www.airis4d.com

Imagine you are talking to a machine that understands you perfectly, like a trusted friend or family

member. You can tell it what you need, and it responds in a way that makes sense, almost like it can read your

mind. This may sound like science ﬁction, but it’s actually becoming a reality, thanks to advances in natural

language technologies like NLP, NLU, and NLG. It is important to note that these acronyms should not be used

interchangeably, as NLP(Natural Language Processing), NLG(Natural Language Generation), and NLU(Natural

Language Understanding) are related ,yet distinct technologies that serve as the building blocks for a machine’s

language skills. These technologies allow machines to understand, interpret, and generate human language. At

a high level, NLU and NLG are just components of NLP. In this article, we will explore the fascinating world of

natural language technologies and how they are bringing us closer to a future where we can communicate with

machines just like we do with each other.

2.1 How NLU and NLG ﬁt into NLP

NLP, a subﬁeld of artiﬁcial intelligence, involves the interaction between computers and human languages.

Within NLP, NLU is responsible for enabling machines to comprehend and interpret human language. NLU

achieves this by breaking down words and sentences into individual components, considering context, and

identifying the meaning behind the words used. NLG, on the other hand, utilizes the information gathered from

NLU to generate a natural-sounding and situation-appropriate response. NLG can even incorporate idioms and

colloquialisms, giving the conversation a more human-like feel. Working together, these components of NLP

allow machines to understand and generate human language(Figure 1)

2.2 Natural Language Understanding

Natural Language Understanding (NLU) refers to a computer’s ability to comprehend and interpret human

language. Before a computer can process unstructured text into a machine-readable format, it needs to understand

the peculiarities of natural language. These complexities include nuances in grammar, syntax, and semantics

that are often challenging for machines to decipher. NLU provides algorithms and models that help computers

analyze and comprehend the meaning of natural language. Without NLU, machines would struggle to process

human language, limiting their ability to extract insights and value from this vast source of information.

2.2 Natural Language Understanding

Figure 1: Relation between NLP, NLU and NLG

(image courtesy:https://grammarist.com/heteronyms/tear-vs-tear/)

Figure 2: Tear vs Tear

For example, consider the following sentence: I saw a tear in her eye as she tried to tear the paper.

In this sentence, the word ”tear” has two diﬀerent meanings. The ﬁrst ”tear” refers to a drop of water coming

out of someone’s eye when they cry, while the second ”tear” means to rip or cut something apart (Figure 2).

The context of the sentence makes it clear which meaning is intended: the tear in her eye indicates that the

ﬁrst meaning of ”tear” is being used, while the fact that she is trying to tear the paper suggests that the second

meaning of ”tear” is intended.

Here, NLU algorithms would analyze the surrounding words and phrases to determine whether the word

”tear” is being used as a verb or a noun. By considering the context in which the word is used, NLU technology

could accurately identify the intended meaning of the word ”tear” in each instance, allowing machines to

understand the subtle nuances of language and interpret the sentence correctly.

2.2.1 Intent recognition and Entity recognition

Two fundamental concepts of NLU are intent and entity recognition.

Intent recognition is a crucial component of NLU (Natural Language Understanding) that involves iden-

tifying the user’s underlying purpose or objective in the input text. This process not only involves identifying

the words and phrases used in the input text, but also interpreting the context and tone to determine the user’s

sentiment. As such, intent recognition is considered the ﬁrst and most important step in NLU, as it sets the

foundation for the subsequent processing and analysis of the text.

Entity recognition involves identifying the speciﬁc pieces of information or entities in a message and

extracting the most important information about them. Entities can be thought of as parameters or variables that

help to deﬁne the context of the user’s input. There are two main types of entities that can be identiﬁed through

entity recognition: named entities and numeric entities. Named entities refer to speciﬁc entities that have a

2.3 Natural Language Generation

(image courtesy:https://www.artefact.com/blog/nlu-benchmark-for-intent-detection-and-named-entity-recognition-in-call-center-conversations/)

Figure 3: llustration of a customer intent and several entities that are extracted from conversation

proper name, such as people, companies, locations, and organizations. Numeric entities, on the other hand, are

entities that are recognized as numbers, currencies, and percentages. By identifying and extracting the entities

in a user’s input, NLU systems can gain a deeper understanding of the user’s needs and provide more relevant

and accurate responses.

To gain a clear understanding of the concepts of intent and entity, let us consider the following sentence.

”Hi, I’m calling to know if the small black bag is available in NYC?” . The intent of the sentence is to

inquire about the availability of a speciﬁc item - the small black bag - in a particular location, which is NYC.

The entity in this sentence is the speciﬁc item being inquired about, which is the ”small black bag.” This is a

named entity that belongs to the category of products or items. Another entity in the sentence is the location,

which is ”NYC” or New York City. This is also a named entity that belongs to the category of locations(Figure

NLU is important for a wide range of applications, including chatbots, virtual assistants, and customer

service systems. By accurately understanding the intent and entities of a user’s input, these systems can provide

more eﬀective and personalized responses.

2.3 Natural Language Generation

Natural language generation (NLG) is the process of generating text that appears to be written by a human,

without the need for a human to actually write it. This means that NLG systems can generate text that sounds like

it was written by a person, rather than a machine. The purpose of NLG is to help computers communicate more

eﬀectively with humans, by making the text they produce more understandable and relatable. Sophisticated

NLG software is capable of analyzing vast amounts of structured or unstructured data and extracting meaningful

insights. By using advanced algorithms and machine learning models, NLG systems can identify patterns,

correlations, and relationships within the data that might not be immediately obvious to human analysts.

Once the data has been analyzed, NLG software can generate human-like language output that summarizes

the key ﬁndings and insights. This text can be customized to suit the needs of the user, and can be presented in

a variety of formats, such as reports, dashboards, or visualizations.

2.3 Natural Language Generation

(image courtesy:https://research.aimultiple.com/nlg/)

Figure 4: Stages of NLG

2.3.1 Stages of NLG

NLG involves a series of stages, with each step reﬁning the data to generate natural-sounding language.

The process consists of six stages(Figure 4), which are:

1.Content Analysis: In this stage, the input data is analyzed to determine what should be included in the

ﬁnal content. For example, an NLG system analyzing a weather forecast might identify the key information,

such as temperature, wind speed, and precipitation, and determine how they are related.

2.Data Understanding: The NLG system then interprets the data and identiﬁes patterns and relationships

using machine learning algorithms, such as recognizing that high humidity levels are often associated with

increased chances of precipitation.

3.Document Structuring: Once the data has been analyzed and understood, the NLG system creates a

plan for the document and chooses a narrative structure based on the type of data being interpreted. For example,

it might decide on a headline, an opening paragraph, and a series of subheadings to structure the forecast.

4.Sentence Aggregation: In this stage, relevant sentences or parts of sentences are combined in ways

that accurately summarize the topic, such as generating a sentence like ”Expect scattered thunderstorms in the

afternoon with temperatures reaching the mid-80s.”

5.Grammatical Structuring: The NLG system applies grammatical rules to generate natural-sounding

text, such as determining the syntactical structure of a sentence and rewriting it in a grammatically correct

manner like ”There is a 40% chance of rain tomorrow.”

6.Language Presentation: The ﬁnal output is generated based on a template or format the user or

programmer has selected. For example, the NLG system might output the information in a visually appealing

and easy-to-read format, such as a table or chart, with icons or symbols representing diﬀerent weather conditions.

In summary, NLP, NLU, and NLG are technologies that are rapidly evolving and gaining importance in

various industries. NLP involves the analysis and manipulation of natural language data, NLU focuses on

machine comprehension and interpretation of human language, and NLG is concerned with the creation of

human-like language by machines. These technologies have the potential to revolutionize the way humans and

machines interact and communicate, leading to greater eﬃciency, accuracy, and automation in various ﬁelds

such as healthcare, ﬁnance, and customer service.

2.3 Natural Language Generation

References

NLP, NLU, and NLG: What’s The Diﬀerence? A Comprehensive Guide, Nahla Davies, KDnuggets, June

2022

What Is Natural Language Understanding (NLU)?, Rachel Wolﬀ, MonkeyLearn, January 2021

NLP vs. NLU vs. NLG: the diﬀerences between three natural language processing concepts, Eda

Kavlakoglu, IBM Blog, November 2020

Natural Language Understanding (NLU), TechTarget

Understanding NLU benchmarks for intent detection and named-entity recognition in call centre conver-

sations, Artefact Data Digest, November 2020

Natural Language Generation (NLG) in 2023, Cem Dilmegan,AI Multiple, January 2023

Natural Language Generation (NLG),Ivy Wigmore, TechTarget

What is Natural Language Generation?, AI, Data & Analytics Network, July 2022

About the Author

Jinsu Ann Mathew is a research scholar in Natural Language Processing and Chemical Informatics.

Her interests include applying basic scientiﬁc research on computational linguistics, practical applications of

human language technology, and interdisciplinary work in computational physics.

Part II

Astronomy and Astrophysics

Exploring the Explosive Side of the Sun

by Linn Abraham

airis4D, Vol.1, No.5, 2023

www.airis4d.com

1.1 The Sun and Life on Earth

Energy is a fundamental requirement for life. The most common characteristics of life, namely, organi-

zation, metabolism, response to stimuli, reproduction, adaptation and evolution, all require energy. All almost

life forms on Earth depend on the Sun for their energy (There are organisms that can derive their energy by

oxidizing inorganic compounds such as hydrogen sulﬁde, ammonia or methane. They are mostly found in

environments without sunlight such as deep-sea hydrothermal vents, hot springs, and caves). It is no wonder

then that almost of all the early civilizations and cultures worshiped the Sun as a deity. The warmth and light

of the Sun was a welcome relief from the cold and dark nights that made him afraid. The Sun’s regular rising

and setting provided a sense of order and predictability to the natural world, which was often seen as chaotic

and unpredictable. The brightness and intensity of the Sun gave it a divine glow that we have come to associate

with many other deities. Without any apparent source of energy the Sun was a mystery to the ancients and this

made it worthy to be worshiped by countless generations of people.

1.2 Historical Record of Solar Explosions

Today, worshiping the Sun might sound to be a ridiculous thing to do. The scientiﬁc revolution brought

about a paradigm shift in our perception of the Sun. The Italian astronomer Galileo Galilei made the ﬁrst

recorded observations of Sunspots. Galileo’s observations showed that the Sun had dark spots on its surface that

appeared to change in size and shape over time. These observations were groundbreaking, as they challenged

the prevailing view of the Sun as a perfect, unblemished celestial body, and paved the way for further research

into the Sun’s physical properties and behavior. Samuel Heinrich Schwabe discovered the sunspot cycle in

1844. This is a periodic variation in the number of sunspots observed. The average period is seen to be close

to 11 years with the actual period between 7 and 17 years. The fact that there can be sudden explosions on

the surface of the sun was discovered on 1 September 1859 by Richard Carrington, an English astronomer.

Carrington was studying a sunspot with his telescope. As is usually the case with astronomers who study the

sun, he projected an image of the sun onto a screen and observed it. He was probably surprised to ﬁnd that ‘two

patches of intensely bright and white light broke out’ in the middle of the sunspot.

1.3 Do We Need to Predict These Outbursts?

Figure 1: Sun Blasts a M6.6 Flare On Feb. 13th at 1738 UT, sunspot 1158 unleashed the strongest solar ﬂare

of the year so far, an M6.6-category blast. Image Credit: NASA/ SDO/ AIA

Figure 2: The Earth is dwarfed by a giant explosion on the Sun that ejects huge amounts of hot gas into space.

Here the sizes of the Earth and the Sun are shown to scale, but not the distance between them. Image Credit:

NASA

1.3 Do We Need to Predict These Outbursts?

About 18 hours after he saw the ﬂare, the magnetic observing station in Kew reported a sudden violent

change in the earth’s magnetic ﬁeld, known as a geomagnetic storm. The storm was so strong that it disrupted

telegraph communications around the world and caused auroras to be visible at lower latitudes than usual. One

of the earliest known geomagnetic storms was the ”Bastille Day Event” of July 14, 1770. This storm was

observed by French astronomer Jean-Andre Deluc, who noted that a compass needle in his possession was

behaving erratically. Other observers in Europe and North America also reported unusual auroras and magnetic

disturbances around the same time. Carrington made the ﬁrst indirect suggestion made by anybody that an

explosion in the sun could cause abnormal things to happen on the Earth. He made this observation at the

meeting of the Royal Astronomical Society about two months after the discovery. Several other solar storms

have been observed ever since. The greatest solar storm of the Space Age was observed in 1989 which led to a

blackout in the Canadian province of Quebec due to the outage of their power grid. Interestingly, the ﬁrst ﬂare

recorded by any human being (dubbed “The Carrington Event”) is also the strongest ﬂare ever to have been

observed. The evidence for this was unearthed by researchers who found measurements of the storm as far away

from the geomagnetic poles as India. This ﬂare must have been at least three times as strong as the 1989 ﬂare

that caused havoc in Quebec.

1.4 How the Modern World is Becoming More Vulnerable to Solar Flares?

The fact that the Carrington event was not recorded as a catastrophic event in history whereas the Quebec

solar ﬂare made major headlines shows how the world has become increasingly dependent on technology. Not

just any technology but technology that is vulnerable to solar ﬂares. The connection between solar ﬂares and

modern technology can be understood with Michael Faraday’s observation of magnetic ﬁelds inducing electric

currents. Our current understanding of the sun shows that it must be a ball of hot burning gas. But the gas is

mostly in the inner layers, whereas the outer layers exists as plasma due to the very high temperatures. In 1958

Eugene Parker proposed the theory of the Solar Wind where he found that the high temperature of the corona

should cause a continuous wind of gas and plasma to blow through the entire solar system. Solar ﬂares cause a

sudden increase in the inﬂux of these particles. The earth’s magnetic ﬁeld mitigate the eﬀects of sudden solar

disturbances to some extent. However this protective shield is weakest near the geomagnetic poles because

of a lower density of magnetic ﬁeld lines. The polar auroras are a harmless consequence of this fact. One

geomagnetic pole is located in the Canadian Arctic. This explains why the major blackout occurred in the

Canadian province of Quebec. The interaction of the electrical charged particles with the Earth’s magnetic ﬁeld

is capable of causing induced currents in our electrical grids and subsequently cause them to fail. There are

several other ways in which solar ﬂares aﬀect our lives. Radio communications are an essential part of modern

technologies. The ionosphere is what makes radio communications to distant places on the Earth possible

which would otherwise be impossible due to the Earth’s curvature. Solar ﬂares disturb the ionosphere and thus

cause disturbances in radio communication. The north geomagnetic pole is a frequent route for passengers

between Europe and North America. After a major solar ﬂare, airline companies would have to reroute ﬂights

else the pilots lose radio contact with ground stations. Artiﬁcial satellites are another piece of technology that

has become ubiquitous to life on Earth. We rely on them for everything from television broadcasting to daily

navigation (GPS satellites). A major solar ﬂare can roast the electronics on board if precautions are not put in

place. All this points to an increased need to understand the dynamics of our Sun. Knowing more about the

solar activity and sunspots are critical. Advance warning of the solar ﬂare predictions are important for disaster

1.5 References

Figure 3: Comet NEOWISE is visible in an aurora-ﬁlled sky in this photo taken on July 14, 2020, by

Aurorasaurus Ambassador Donna Lach. Image Credit: NASA

mitigation.

1.5 References

1. Choudhuri, Arnab Rai, Nature’s Third Cycle: A Story of Sunspots, Oxford, New York, Oxford University

Press, 2015.

2. Karttunen, Hannu, Pekka Kr

oger, Heikki Oja, Markku Poutanen, and Karl Johan Donner, eds. Funda-

mental Astronomy. Berlin, Heidelberg: Springer Berlin Heidelberg, 2017. https://doi.org/10.1007/978-

3-662-53045-0.

3. Did you know that? 12 interesting facts about the Sun.

4. List of solar storms

About the Author

Linn Abraham is a researcher in Physics, specializing in A.I. applications to astronomy. He is

currently involved in the development of CNN based Computer Vision tools for classiﬁcations of astronomical

sources from PanSTARRS optical images. He has used data from a several large astronomical surveys including

SDSS, CRTS, ZTF and PanSTARRS for his research.

Cataclysmic Variable Stars

by Sindhu G

airis4D, Vol.1, No.5, 2023

www.airis4d.com

2.1 What are Cataclysmic Variable stars

Variable stars have been studied by astronomers for many years, and they provide important information

about the nature and behavior of stars, as well as their evolution and physical properties. We already know that

a variable star is a star whose brightness, as observed from Earth, changes over time. This article explains what

cataclysmic variable stars(Figure 1) are. A cataclysmic variable star, often abbreviated as CV star, is a type of

variable star that undergoes sudden and dramatic changes in their properties, leading to signiﬁcant increases in

brightness and energy output. Cataclysmic variables are one type of intrinsic variable stars. Intrinsic variables

are types of variable stars where the variability is caused by changes in the physical properties of the stars

themselves.

Figure 1: Cataclysmic Variable Stars (Image Courtesy: MIT)

Cataclysmic variable (CV) stars are binary systems that typically consist of a normal star (often referred to

as the secondary or donor star) and a white dwarf (the primary star). The primary star in a cataclysmic variable

is typically a white dwarf, which is the remnant of a star that has exhausted its nuclear fuel and has collapsed

to a small, dense state. The companion star can be a main-sequence star, a subgiant, or even another white

dwarf. In cataclysmic variable stars, the gravitational pull of the white dwarf is so strong that it distorts the

shape of the companion star. This distortion can cause the companion star to transfer material onto the white

2.2 Discovery

dwarf through accretion, where material from the companion star forms an accretion disk around the white

dwarf. The accretion disk is a swirling disk of gas and dust that forms around the white dwarf as material

from the companion star accumulates and accretes onto the white dwarf’s surface. The study of mass transfer

and accretion in cataclysmic variables provides valuable insights into the physics of binary star systems and the

evolution of compact objects such as white dwarfs.

Figure 2: Cataclysmic Variable Stars (Image Courtesy: NASA)

In Figure 2, the companion star is shown on the left, with its mass transferring material onto the accretion

disk around the white dwarf (shown on the right). The hot spot is the region where the material from the

companion star impacts the accretion disk, creating a bright spot in the system. The white dwarf is shown at

the center. The interactions between the companion star, the accretion disk, and the white dwarf in cataclysmic

variable stars are complex and dynamic, leading to the observed variability and outbursts in their brightness.

Due to the high density of the white dwarf, the gravitational potential energy released during the accretion

process is enormous. Some of this energy is converted into X-rays, during the accretion process. This X-ray

emission is one of the ways cataclysmic variable stars can be detected and studied, as it provides important

information about the accretion process, the properties of the accretion disk, and the physical conditions in the

vicinity of the white dwarf. Cataclysmic variable stars are generally faint in X-rays. As a result, only cataclysmic

variables that are located relatively close to our Sun, within a few hundred light-years, have been studied in

X-rays so far, while there may be millions of cataclysmic variables in our galaxy. Studying cataclysmic variables

in X-rays provides valuable insights into the high-energy processes and dynamics occurring in these systems,

shedding light on their complex behavior and evolution.

Cataclysmic variable stars are typically small in size, with the entire binary system being comparable in

size to the Earth-Moon system. The orbital periods of cataclysmic variables are usually short, ranging from 1 to

10 hours. This means that the two stars in the binary system are relatively close to each other and orbit around

their common center of mass at a rapid pace.Cataclysmic variable stars are fascinating objects that continue to

be actively researched by astronomers to better understand their properties and evolution.

2.2 Discovery

Cataclysmic variables are indeed a class of astronomical objects that are commonly observed by amateur

astronomers. Cataclysmic variables are known for their irregular and sudden changes in brightness, particularly

2.3 Some types of Cataclysmic Variables

during outbursts, which can make them easily detectable even with modest amateur telescopes. Cataclysmic

variables often have short orbital periods, which means that their brightness can change signiﬁcantly within

hours or days, providing an exciting and dynamic observing experience for amateur astronomers. Cataclysmic

variables are often easily detectable by amateur astronomers due to their distinctive characteristics. They are

typically blue in color, exhibit rapid and strong variability in their brightness, and often have peculiar emission

lines in their spectra. These features, along with their ultraviolet and X-ray emissions, make them relatively

straightforward to identify compared to other celestial objects.

In terms of occurrence, around six galactic novae, which are a type of cataclysmic variable, are discovered

in our own galaxy each year. However, observations in other galaxies suggest that the actual occurrence rate

could be higher, between 20 and 50 per year. The discrepancy between the observed and predicted occurrence

rates could be due to factors such as interstellar dust obscuration, lack of observers in certain regions of the sky

(such as the southern hemisphere), and challenges in observing during daylight or full moon periods.

Amateur astronomers play a valuable role in monitoring cataclysmic variables and contributing to our

understanding of these objects. Their observations, when combined with data from professional observatories,

can help uncover important information about the behavior, properties, and evolution of cataclysmic variables,

as well as shed light on their occurrence rates and other astrophysical processes associated with these intriguing

celestial objects.

2.3 Some types of Cataclysmic Variables

2.3.1 Supernovae

A supernova (Figure 3 and Figure 4) is a cataclysmic stage in the life of a star, characterized by a sudden

and dramatic increase in brightness. Supernovae are one of the most energetic and dramatic events in the

universe. The increase in brightness can be signiﬁcant, with a typical supernova becoming many magnitudes

brighter, often reaching an absolute magnitude of around -15 or even brighter. There are several ways in which

a star can undergo a supernova, including the core collapse of a massive star or the runaway nuclear fusion

in a white dwarf. In the case of a massive star, when it has exhausted the nuclear fuel in its core, the core

collapses under gravity, resulting in an intense release of energy in the form of a supernova explosion. The

outer layers of the star are expelled into space at high velocities, and the core may either collapse to form a

dense neutron star or, if the core is massive enough, it may collapse further to form a black hole. In the case

of a white dwarf, which is the remnant core of a low to medium mass star that has exhausted its nuclear fuel,

a supernova can occur if it accretes enough mass from a companion star to trigger runaway nuclear fusion.

This can result in a thermonuclear explosion that completely destroys the white dwarf, releasing a tremendous

amount of energy in the process. The peak luminosity of a supernova can be incredibly bright, comparable to

the brightness of an entire galaxy, and it can outshine its host galaxy for a short period of time. However, the

brightness of a supernova fades over weeks or months as the ejected material expands and cools. Supernovae

are important astronomical events that provide insights into the processes of stellar evolution, nucleosynthesis,

and the dynamics of the universe. The remnants of supernovae, such as neutron stars and black holes, can also

play signiﬁcant roles in the evolution of galaxies and the universe at large.

2.3 Some types of Cataclysmic Variables

Figure 3: Oldest Recorded Supernova (Image Courtesy: X-ray: NASA/CXC/SAO & ESA; Infared: NASA/JPL-

Caltech/B. Williams (NCSU))

Figure 4: Supernova 1987A in the Large Magellanic Cloud. (Image Courtesy: Anglo-Australian Observatory

2.3.2 Luminous red nova

Luminous red novae (Figure 5) are a type of stellar explosion that is caused by the merger of two stars, and

they are distinct from classical novae. Luminous red novae are known for their characteristic red appearance

and slow decline following the initial outburst.

2.3 Some types of Cataclysmic Variables

Figure 5: V838 Monocerotis – a possible luminous red nova (Image Courtesy: NASA/ ESA)

2.3.3 Novae

A nova(Figure 6) occurs in a close binary system. Novae are known for their rapid and often unpredictable

increase in brightness. The typical range of brightness increase, or ”nova outburst,” is generally between 7 and

16 magnitudes. The behavior of novae, including the rapid rise in brightness followed by a steady decline back

to the pre-nova magnitude over a period of weeks to months, suggests that the event causing the nova does not

typically result in the destruction of the original star. This is consistent with the widely accepted model for

novae, which involves an accreting white dwarf in a close binary system. In this model, the white dwarf accretes

material from its companion star over a long period of time, ranging from thousands to hundreds of thousands of

years, until there is suﬃcient material to trigger a thermonuclear explosion that then blasts the shell of material

oﬀ into space.

Figure 6: Nova KT Eridani(Image Courtesy: Kevin Heider)

2.3.4 Recurrent Novae

Recurrent novae (Figure 7) are similar to novae in that they exhibit outbursts with a change in magnitude

(brightness) typically ranging from 7 to 16 magnitudes, and the period of outburst can be up to about 200 days.

Recurrent novae are characterized by multiple outbursts over recorded observations.

2.3 Some types of Cataclysmic Variables

Figure 7: Recurrent Novae, T Pyxidis Light Echoes(Image Courtesy: NASA/ESA)

2.3.5 Dwarf novae

Figure 8: Dwarf Novae, Z Camelopardalis(Image Courtesy: NASA/JPL-Caltech/T. Pyle(SSC)/R. Hurt(SSC))

Dwarf novae (Figure 8) are a type of variable star that exhibit sudden and signiﬁcant increases in brightness

by 2 to 5 magnitudes over a few days, with intervals of weeks or months between outbursts. They are intrinsically

faint stars and are characterized by a binary system with a white dwarf as one of the component stars. There are

three subtypes of dwarf nova that have been identiﬁed: U Geminorum, Z Camelopardalis, and SU Ursae Majoris

stars. The most widely accepted model to explain the outbursts of dwarf novae is the disk instability model.

According to this model, the accretion disk around the white dwarf undergoes thermal instabilities, leading

2.3 Some types of Cataclysmic Variables

to sudden increases in brightness during outbursts. Unlike classical novae, there is no signiﬁcant ejection of

material in dwarf novae events.

References

Understanding Variable Stars, John R Percy, Cambridge University Press.

GCVS variability types, Samus N.N., Kazarovets E.V., Durlevich O.V., Kireeva N.N., Pastukhova

E.N.,General Catalogue of Variable Stars: Version GCVS 5.1,Astronomy Reports, 2017, vol. 61, No. 1,

pp. 80-88 2017ARep...61...80S

Cataclysmic variable star

Cataclysmic Variables

Variable star

About the Author

Sindhu G is a research scholar in Physics doing research in Astronomy & Astrophysics. Her research mainly

focuses on classiﬁcation of variable stars using diﬀerent machine learning algorithms. She is also doing the

period prediction of diﬀerent types of variable stars, especially eclipsing binaries and on the study of optical

counterparts of x-ray binaries.

Part III

Biosciences

Barcode of Life: A Global Biodiversity

Challenge

by Geetha Paul

airis4D, Vol.1, No.5, 2023

www.airis4d.com

Biodiversity, the variety of life on Earth, is fundamental to the well-being of our planet and all its inhabitants.

However, the rapid pace of human development and the resulting destruction of natural habitats have led to a

dramatic decline in biodiversity over the past few decades. In response, scientists and conservationists around

the world have been working to document and protect the planet’s species before they disappear forever. One of

the most promising initiatives in this eﬀort is the ‘Barcode of Life’ (BOLD) project.

The Barcode of Life project is a global biodiversity challenge that aims to create a curated, comprehensive,

digital library of DNA barcodes for every species on Earth. DNA barcoding involves analysing a speciﬁc region

of an organism’s DNA to identify it with a unique code, similar to a product barcode. This code can then be

used to quickly and accurately identify the species, even when only a small fragment or specimen is available.

The BOLD project was launched in 2003 by Paul Hebert, a biologist at the University of Guelph in Canada.

Since then, it has grown to include hundreds of institutions and thousands of researchers from around the world.

The project has already generated over 10 million DNA barcode records, representing more than 260,000

species.

The potential applications of the Barcode of Life project are vast. DNA barcoding can be used to identify

species in a wide range of ﬁelds, including ecology, agriculture, medicine and also in food safety & quality. For

example, it can be used to identify the species of ﬁsh being sold in a market, can determine which insects are

responsible for crop damage, or to diagnose infectious diseases. In addition, the Barcode of Life project can

help in conservation eﬀorts by enabling researchers to identify and monitor threatened and endangered species

more eﬃciently.

The BOLD project also serves as a platform for collaboration among scientists, conservationists, and

policymakers from around the world. By bringing together experts from diﬀerent ﬁelds, the project fosters

interdisciplinary research and helps to bridge gaps in knowledge and resources.

However, there are also challenges associated with the Barcode of Life project. One of the main challenges

is the sheer scale of the project. With an estimated 8.7 million species on Earth, creating a complete DNA

barcode library is a monumental task. In addition, there are technical and logistical challenges associated with

collecting and analysing DNA samples from such a diverse range of organisms. Despite these challenges,

the Barcode of Life project has already made signiﬁcant progress in documenting and preserving the world’s

biodiversity. As the project continues to grow, it has the potential to revolutionise the way we understand and

protect life on Earth. The Barcode of Life follows a standardised procedure for generating DNA barcodes, which

includes the following steps:

Figure 1: Diagram showing the steps of DNA barcoding coupled with next-

generation sequencing (NGS) for determination of biodiversity composition in

complex environmental samples.Image courtesy:https://www.researchgate.net/ﬁgure/

a-Diagram-showing-the-steps-of-DNA-barcoding-coupled-with-next-generation-sequening-NGS

Figure 2: Schematic drawing of the steps involved in the creation of reference libraries of DNA barcodes.

Note the links with involved institutions such as natural history museums, and related initiatives like the

Encyclopedia of Life. (With permission from the Consortium for the Barcode of Life) Image courtesy:https:

//www.mdpi.com/1424-2818/13/7/313

The various steps involved in the barcoding pipeline are as follows:

1. Specimen collection and identiﬁcation: Collecting a specimen from the ﬁeld and identifying it to the

species level is the ﬁrst step in generating a DNA barcode. Specimens are identiﬁed using traditional

taxonomic methods such as morphology studies.

2. DNA extraction: DNA is extracted from the collected specimen using standard laboratory techniques. The

DNA extraction protocol depends on the nature of the specimen, the type of tissue, and the preservation

method used.

3. PCR ampliﬁcation of barcode region: The barcode region is ampliﬁed using the polymerase chain reaction

(PCR) with universal primers designed to target the barcode region of interest. The barcode region is

typically a short fragment of a standard mitochondrial gene, such as cytochrome c oxidase I (COI) in

animals.

4. Sequencing: The PCR products are sequenced using Sanger sequencing or next-generation sequencing

(NGS) technologies. Sanger sequencing is the traditional method of sequencing and can generate high-

quality, long reads, while NGS provides high-throughput sequencing of short reads. The PCR products

are prepared into a library that can be sequenced using NGS technologies such as Illumina, PacBio,

or Oxford Nanopore. The prepared library contains many diﬀerent barcode sequences from diﬀerent

specimens.

5. Data analysis and quality control: The NGS data generated from the raw sequencing data are processed

and analysed using bioinformatics tools to obtain the DNA barcode sequence,to identify the species

present in the sample. The sequence is then compared to a reference library to conﬁrm the species

identiﬁcation. Quality control measures are applied to ensure the accuracy and reliability of the barcode

sequence.

6. Interpretation of results: Finally, the results are interpreted to understand the biodiversity composition of

the sample. The number of unique sequences identiﬁed in the sample can be used to estimate the species

richness, while the abundance of each sequence can be used to estimate the relative abundance of each

species.

7. Data submission and storage: The generated barcode sequence and associated metadata are submitted

1.1 NEXT GENERATION SEQUENCING (NGS)

to a centralised database, such as the Barcode of Life Data Systems (BOLD) or GenBank, for long-term

storage and public access.

The Barcode of Life procedure provides a standardised, cost-eﬀective, and eﬃcient method for generating

DNA barcodes for all eukaryotic species. It has many applications in various ﬁelds, including biodiversity

conservation, ecology, forensics, medicine and biosecurity.

Overall, DNA barcoding coupled with Next Generation Sequencing (NGS) provides a powerful tool for

studying biodiversity composition in complex environmental samples. Billions of DNA strands get sequenced

simultaneously using NGS. Whereas with Sanger Sequencing, only one strand is sequenced at a time. While

the Human Genome Project took over 30 years to sequence the human genome for the ﬁrst time. Now with Next

Generation Sequencing, a whole human genome can be sequenced in just one day. NGS only works because

the Human Genome Project created a human reference DNA sequence. The basic principle behind NGS is that

DNA can be cut into smaller pieces and sequenced. Next Generation Sequencing (NGS) is used to sequence

both DNA and RNA. First, samples get collected, and the DNA or RNA gets puriﬁed. Next, the DNA or

RNA gets checked to ensure its purity. RNA ﬁrst needs to be reverse-transcribed into DNA before it can get

sequenced. A library then gets prepared from the DNA. A library is the collection of short DNA fragments

from a long stretch of DNA. Libraries get made by cutting the DNA into short pieces of a speciﬁc size. This

cutting gets done using high-frequency sound waves or enzymes. Then sequences of DNA called adapters get

added to each end of a DNA fragment. These adapters contain the information needed for sequencing. They

also include an index to identify the sample. Finally, any non-bound adapters get removed, and the library is

complete. Depending on the application, there can be a PCR (Polymerised Chain Reaction) step to increase the

library amount.

1.1 NEXT GENERATION SEQUENCING (NGS)

Figure 3: The ﬁgure shows the Libraries obtained by cutting the DNA into short pieces of a speciﬁc size. Image

courtesy:https://www.clevalab.com/post/a-step-by-step-guide-to-ngs

A successful library will be of the correct size. It will also be of a high enough concentration for sequencing.

Next step is to attach the library to FlowCell.

1.1 NEXT GENERATION SEQUENCING (NGS)

Figure 4: The main sequencing instruments used in NGS are from Illumina. These instru-

ments use a method called sequencing by synthesis.Image courtesy:https://www.ibiology.org/techniques/

next-generation-sequencing

Figure 5: Attaching the library to the ﬂow Cell - sequencing by synthesis. Image courtesy:https://www.clevalab.

com/post/a-step-by-step-guide-to-ngs

The sequencing occurs on a glass surface of a ﬂow cell. Short pieces of DNA called oligonucleotides are

bound to the surface of the ﬂow cell. These oligonucleotides match the adapter sequences of the library. First,

the library gets denatured to form single DNA strands. Then this library gets added to the ﬂow cell, which

attaches to one of the two oligos. The strand that attaches to the oligo is the forward strand. Next, the reverse

strand gets made, and the forward strand gets washed away. The library is now bound to the ﬂow cell.

If sequencing started now, the ﬂuorescent signal would be too low for detection. So each unique library

fragment needs to get ampliﬁed to form clusters. This Clonal Ampliﬁcation is by a PCR that happens at a single

temperature. Annealing, extension and melting occur by changing the ﬂow cell solution. First, the strands

bind to the second oligo on the ﬂow cell to form a bridge. The strands get copied. Then these double-stranded

1.1 NEXT GENERATION SEQUENCING (NGS)

fragments get denatured. This denaturation gets done by adding another solution to the ﬂow cell. This copying

and denaturing repeats over and over. Localised clusters get made, and ﬁnally, the reverse strands get cut. These

strands get washed away, leaving the forward strand ready for sequencing.

Figure 6: Clonal ampliﬁcation where localised clusters are made. Image courtesy: https://www.clevalab.com/

post/a-step-by-step-guide-to-ngs

Figure 7: Sequencing of the DNA strand and addition of ﬂuorescent G,C,T,A nucleotides. Image courtesy:

https://www.clevalab.com/post/a-step-by-step-guide-to-ngs

The sequencing primer binds to the forward strands. Next, ﬂuorescent nucleotides G, C, T and A get

added to the ﬂow cell along with DNA Polymerase. Each nucleotide has a diﬀerent colour ﬂuorescent tag

and a terminator. So only one nucleotide can get sequenced at a time. First, the complementary base binds

to the sequence. Then the camera reads and records the colour of each cluster. Next, a new solution ﬂows in

and removes the terminators. The nucleotides and DNA Polymerase ﬂow in again, and another nucleotide gets

sequenced. These read cycles continue for the number of reads set on the sequencer. Once complete, these

read sequences get washed away. Then the ﬁrst index gets sequenced, then washed away. If only a single read

is needed, the sequencing ends here. But, for paired-end sequencing, the second index is sequenced, as well as

the reverse strand of the library. There is no primer for the second index read. Instead, a bridge gets created so

that the second oligo acts as the primer. The second index is then sequenced. These two index reads use unique

dual indexes. These allow the use of up to 384 samples in the same ﬂow cell. Next, the reverse stands get made,

1.1 NEXT GENERATION SEQUENCING (NGS)

and the forward strands are cut and washed away. The reverse stands are then sequenced.

Figure 8: Filtering of bad reads. Image courtesy: https://www.clevalab.com/post/a-step-by-step-guide-to-ngs

Figure 9: Aligning and Making Sense of the Reads. Image courtesy: https://www.clevalab.com/post/

a-step-by-step-guide-to-ngs

Once the sequencing is complete, any bad reads get ﬁltered out. These include the clusters that overlap,

lead or lag with sequencing or are of low intensity. The clusters can’t overlap on a patterned ﬂow cell, but there

can be more than one library fragment per nanowell. These polyclonal wells will also get ﬁltered out. Next,

the reads passing the ﬁlter get demultiplexed. Demultiplexing uses the attached indexes to identify and sort

reads from each sample. Finally, the reads get mapped to the reference genome. The diﬀerent reads align to the

reference genome, overlapping each other. Paired-end sequencing creates two sequencing reads from the same

library fragment. During sequence alignment, the algorithm knows that these reads belong together. Longer

stretches of DNA or RNA can get analysed with greater conﬁdence that the alignment is correct.

Next step is the Aligning and Making Sense of the Reads

Read depth is an essential metric in sequencing. Read depth is the number of reads for a nucleotide.

Average read depth is the average depth across the region sequenced. For whole genome sequencing, a 30x

1.2 DNA barcoding as a new tool for food traceability

average read depth is good. A 1500x average read depth is suitable for detecting rare mutation events in cancers.

Another essential metric is coverage. The aim is to have no missing areas across the target DNA.

1.2 DNA barcoding as a new tool for food traceability

Food safety and quality are nowadays a major concern. Any case of food alteration, especially when

reported by the media, has a great impact on public opinion. There is an increasing demand for the improvement

of quality controls, hence addressing scientiﬁc research towards the development of reliable molecular tools

for food analysis. DNA barcoding is a widely used molecular-based system, which can identify biological

specimens, and is used for the identiﬁcation of both raw materials and processed food. In this review the results

of several researches are critically analysed, in order to exploit the eﬀectiveness of DNA barcoding in food

traceability, and to delineate some best practices in the application of DNA barcoding throughout the industrial

pipeline. The use of DNA barcoding for food safety and in the identiﬁcation of commercial fraud is also

discussed.

1.3 Highlights

▶ Food quality: from the ﬁeld to the table. ▶ From molecular-based approaches to DNA barcoding. ▶

DNA barcoding to identify and certify food raw material. ▶ DNA barcoding as a traceability tool during food

industrial processing. ▶ Food safety and commercial frauds.

1.4 Strengths, Weaknesses, Opportunities, and Threats for DNA barcoding,

resulting from the SWOT analysis.

Figure 10: Major Strengths, Weaknesses, Opportunities, and Threats for DNA barcoding, resulting from the

SWOT analysis. Image courtesy: https://www.mdpi.com/1424-2818/13/7/313

In conclusion, the Barcode of Life project is a global biodiversity challenge with the ambitious goal of

creating a comprehensive, digital library of DNA barcodes for every species on Earth. By providing a rapid

1.4 Strengths, Weaknesses, Opportunities, and Threats for DNA barcoding, resulting from the SWOT analysis.

and accurate means of species identiﬁcation, the project has the potential to transform our understanding of the

planet’s biodiversity and support conservation eﬀorts worldwide. While there are challenges associated with the

project, the Barcode of Life initiative represents an important step forward in our eﬀorts to protect the natural

world.

References

Biological identiﬁcations through DNA barcodes; Paul D N Hebert Alina Cywinska, Shelley L Ball,

Jeremy R deWaard,. 2003 Feb 7;270(1512):313-21. doi: 10.1098/rspb.2002.2218. https://pubmed.ncbi.

nlm.nih.gov/12614582/

DNA barcoding as a new tool for food traceability; Andrea Galimberti , Fabrizio De Mattia , Alessia Losa

, Ilaria Bruni, Silvia Federici , Maurizio Casiraghi , Stefan Martellos , Massimo Labra; Food Research

International, Volume 50, Issue 1, January 2013, Pages 55-63 https://doi.org/10.1016/j.foodres.2012.09.

036

https://www.researchgate.net/publication/228231851 The Contribution of the Barcode of Life Initiative

to the Discovery and Monitoring of Biodiversity

Grant, D.M.; Brodnicke, O.B.; Evankow, A.M.; Ferreira, A.O.; Fontes, J.T.; Hansen, A.K.; Jensen, M.R.;

Kalaycı, T.E.; Leeper, A.; Patil, S.K.; Prati, S.; Reunamo, A.; Roberts, A.J.; Shigdel, R.; Tyukosova, V.;

Bendiksby, M.; Blaalid, R.; Costa, F.O.; Hollingsworth, P.M.; Stur, E.; Ekrem, T. The Future of DNA

Barcoding: Reﬂections from Early Career Researchers. Diversity 2021, 13, 313.https://www.mdpi.com/

1424-2818/13/7/313

About the Author

Geetha Paul is one of the directors of airis4D. She leads the Biosciences Division. Her research

interests extends from Cell & Molecular Biology to Environmental Sciences, Odonatology, and Aquatic Biology.

Part IV

Climate

Tropical Cyclones

by Robin Jacob Roy

airis4D, Vol.1, No.5, 2023

www.airis4d.com

A tropical cyclone is a rapidly rotating powerful weather system characterized by a low-pressure center, a

closed low-level atmospheric circulation, strong winds, and a spiral arrangement of thunderstorms that produce

heavy rain and squalls. Tropical cyclones pose a major threat to both human life and property, even during

their early stages of development. These weather systems are composed of various hazards that can each cause

signiﬁcant impacts on their own, including storm surges, ﬂooding, powerful winds, tornadoes, and lightning.

When these hazards combine and interact with one another, the potential for loss of life and damage to property

increases substantially. As a result, it is important to understand and prepare for the dangers posed by tropical

cyclones. Figure 1 shows the image of Hurricane Ian, a powerful Category 5 Atlantic hurricane which strike

the state of Florida since the 1935 Labor Day hurricane.

1.1 Formation

Tropical cyclones are powered by the heat and moisture from the warm ocean surface, which causes the air

to rise rapidly, creating an area of low pressure. As the warm air rises, it cools and condenses, forming clouds

and releasing heat, which fuels the storm’s growth. The rotation of the Earth then causes the storm to spin, with

winds picking up speed as it intensiﬁes. Once a tropical cyclone reaches sustained winds of at least 74 mph

(119 km/h), it is classiﬁed as a hurricane or typhoon, depending on the region where it forms. These storms can

cause signiﬁcant damage to coastal communities and are closely monitored by meteorologists around the world.

Before a tropical cyclone can develop, various favorable environmental factors need to be present. These

include:

Warm ocean waters (at least 27°C) throughout a depth of about 46 m.

An atmosphere which cools fast enough with height such that it is potentially unstable to moist convection.

Relatively moist air near the mid-level of the troposphere (4,900 m).

Generally, a minimum distance of at least 480 km from the equator.

A pre-existing near-surface disturbance.

Vertical wind shear, which refers to the diﬀerence in wind speed at various altitudes, should be minimal

(less than approximately 37 km/h) from the surface to the upper troposphere for a tropical cyclone to

form.

1.2 Structure

Figure 1: Hurricane Ian pictured from the International Space Station as it orbited 415 kms above the Caribbean

Sea east of Belize on September 26, 2022. Source: NASA.

Figure 2: Cross-section of a tropical cyclone in the Northern Hemisphere Source: Wikimedia Commons.

1.2 Structure

Tropical cyclones are complex, multi-level, and multi-dimensional systems characterized by strong winds,

heavy rainfall, and low atmospheric pressure. The structure of a tropical cyclone can be divided into several

parts, each with its own characteristics and functions.

1. Eye: The center of a tropical cyclone is called the ”eye.” It is an area of low pressure, where the winds

are light, and the skies are clear or partly cloudy. The size of the eye varies from a few kilometers to more

than 50 kilometers in diameter in the most intense cyclones. The pressure in the eye is typically around

10 percent lower than the pressure outside the cyclone.

2. Eyewall: The eyewall is the area surrounding the eye, where the strongest winds and heaviest rainfall are

found. The eyewall is where the most intense thunderstorms occur, with updrafts and downdrafts that can

reach over 20 kilometers in height. The eyewall is the most dangerous part of a tropical cyclone, where

the winds can exceed 200 kilometers per hour and the rainfall can be more than 100 millimeters per hour.

3. Spiral Rainbands: The spiral rainbands extend outward from the eyewall and are responsible for much

of the rainfall associated with tropical cyclones. They are curved bands of clouds that rotate around the

1.3 Classiﬁcation

center of the storm. These bands can be hundreds of kilometers long and tens of kilometers wide. The

rainfall within these bands can cause ﬂash ﬂooding and landslides.

4. Upper-level outﬂow: Tropical cyclones draw in warm, moist air from the surface, which rises and cools as

it reaches higher altitudes. This rising air creates a low-pressure area at the surface, which draws in more

warm, moist air. The air that rises within the cyclone spreads out at the upper levels of the atmosphere,

forming a high-pressure area, called the ”upper-level outﬂow.” This outﬂow helps to ventilate the cyclone

and maintain its circulation.

5. Sea surface: Tropical cyclones develop over warm ocean waters, where the sea surface temperature is

typically above 26.5°C. The warm water provides the energy for the cyclone to form and intensify, as it

evaporates and rises to form clouds and thunderstorms. The strong winds associated with the cyclone

also stir up the ocean, causing large waves and dangerous storm surges.

Figure 2 illustrates a cross-section of a typical hurricane that occurs in the Northern Hemisphere. Overall,

the structure of a tropical cyclone is a complex and dynamic system, with multiple components interacting with

each other to maintain its circulation and intensity. The interaction between the diﬀerent parts of the cyclone is

critical to its development and behavior, and understanding this interaction is essential for predicting the track

and intensity of the storm.

1.3 Classiﬁcation

Around the world, tropical cyclones are classiﬁed in diﬀerent ways, based on the location (tropical cyclone

basins), the structure of the system and its intensity. For example, tropical storms that occur in the Caribbean

Sea, Gulf of Mexico, North Atlantic Ocean, and the eastern and central North Paciﬁc Ocean are referred to

as ”hurricanes.” In the western North Paciﬁc, they are called ”typhoons.” Meanwhile, in the Bay of Bengal

and Arabian Sea, they are known as ”cyclones.” In the western South Paciﬁc and southeast Indian Ocean, they

are referred to as ”severe tropical cyclones,” while in the southwest Indian Ocean, they are called ”tropical

cyclones.”

The Saﬃr-Simpson Hurricane Wind Scale is a 1 to 5 categorization based on the hurricane’s intensity at

the indicated time. It is a widely recognized system used to measure the strength of hurricanes in the Atlantic

and northern Paciﬁc Oceans. The scale was ﬁrst introduced by Herbert Saﬃr and Robert Simpson in 1971 and

later revised in 2012.

The scale measures hurricanes based on their sustained wind speed, and it is divided into ﬁve categories.

Category 1 hurricanes have sustained wind speeds of 74-95 mph, while Category 5 hurricanes have sustained

wind speeds of 157 mph or higher. Categories 2, 3, and 4 fall in between these two extremes.

In addition to wind speed, the scale also considers the potential damage that a hurricane can cause. For

example, Category 3 hurricanes can cause signiﬁcant damage to homes, buildings, and infrastructure, while

Category 5 hurricanes can cause catastrophic damage, with widespread power outages, structural damage, and

loss of life.

The Saﬃr-Simpson Hurricane Wind Scale is an essential tool for predicting and preparing for hurricane

events. It helps emergency managers and communities to understand the potential impact of a hurricane and to

make decisions about evacuations, emergency response, and other measures to protect life and property.

1.4 Naming

Tropical cyclones have the potential to last for more than a week, leading to the possibility of multiple

cyclones occurring simultaneously. To prevent confusion, each cyclone is assigned a name by weather fore-

casters. Naming conventions for tropical cyclones vary by region, with the Atlantic and Southern hemisphere

(Indian Ocean and South Paciﬁc) using alphabetical order to alternate between male and female names. In the

Northern Indian Ocean, a new naming system was implemented in 2000, with names listed in alphabetical order

by country and using neutral gender. Typically, the list of names is proposed by National Meteorological and

Hydrological Services (NMHSs) of the World Meteorological Organization (WMO) members within a speciﬁc

region and approved by the respective tropical cyclone regional bodies during annual or biennial sessions.

Naming storms, or tropical cyclones, has been a longstanding practice aimed at quickly identifying storms

in warning messages. Using names is believed to be easier to remember than technical terms or numbers, and can

increase community preparedness and interest in storm warnings. The use of short, distinctive names in written

and spoken communications has proven to be quicker and less prone to error than older identiﬁcation methods.

Initially, storms were named arbitrarily, but later on, feminine names were used for storms in the mid-1900s,

and a more organized system was adopted using alphabetically arranged names. Since 1953, Atlantic tropical

storms have been named from lists originated by the National Hurricane Center and updated by an international

committee of the World Meteorological Organization. Lists initially featured only women’s names, but men’s

names were introduced in 1979 and are alternated with women’s names. The lists are rotated, with each list

used again after six years. In the event that a storm’s name is deemed inappropriate due to its impact, it is

stricken from the list, and a new name is selected to replace it. Examples of infamous storm names that have

been stricken from the list include Mangkhut, Irma, Maria, Haiyan, Sandy, Katrina, Mitch, and Tracy.

1.5 Forecasting

Meteorologists worldwide utilize modern technology like weather radars, satellites, and computers to

track tropical cyclones from their development stage. Forecasting tropical cyclones could be challenging

because of their sudden course changes and weakening. Nonetheless, advanced technologies like numerical

weather prediction models help meteorologists forecast a tropical cyclone’s intensity, movement, speed, and

where and when it might hit land. National Meteorological Services of the countries involved release oﬃcial

warnings based on these forecasts. The WMO Tropical Cyclone Programme provides information on these

hazards, while the WMO Severe Weather Information Centre gives real-time tropical cyclone advisories. The

WMO framework enables the timely and widespread dissemination of information about tropical cyclones.

International cooperation and coordination help monitor tropical cyclones from their initial stage of formation.

The WMO coordinates these activities through its Tropical Cyclone Programme. WMO-designated Regional

Specialized Meteorological Centres with activity specialization in tropical cyclones and Tropical Cyclone

Warning Centres detect, track, monitor and forecast all tropical cyclones in their respective regions. These

centres oﬀer advisory information and guidance to National Meteorological and Hydrological Services in

real-time.

References:

Tropical Cyclones

Tropical Cyclone Naming

1.5 Forecasting

Saﬃr-Simpson Hurricane Wind Scale

Mapping the eﬀects of Hurricanes and Cyclones.

About the Author

Robin is a researcher in Physics specializing in the applications of Machine Learning for Remote

Sensing. He is particularly interested in using Computer Vision to address challenges in the ﬁelds of Biodiversity,

Protein studies, and Astronomy. He is currently working on classifying satellite images with Landsat and Sentinel

data.

Part V

General

Bringing Industrial Practices to Academia

by Arun Aniyan

airis4D, Vol.1, No.5, 2023

www.airis4d.com

1.1 Introduction

Workstyle and work culture deeply impact an individual’s personal life. Workstyle is a major factor that

decides the number of fruitful outcomes that one can generate over a period of time. Generally known as

throughput, it also decides the personal satisfaction of one’s work and also drives the path to professional

growth.

In academia, one’s throughput and eﬃciency are measured by the number of publications with high impact

generated over a period of a year. This cannot be fully achieved by one’s individual eﬀort these days. Therefore,

people work on multiple projects in a collaborative fashion which helps in generating more (shared) outputs

and publications. This is a successful model but also has a lot of dependencies when the collaborators are

involved in multiple projects. Even though there is a leader for a collaborative project, timely execution is often

a challenge in academia. The Ph.D. Comics cartoon shown in Figure 1 is a good example of similar situations

when one has his foot in multiple projects and other engagements.

Figure 1: PhD Comics [Image Credit : phdcomics.com]

The work pace in academia, especially with research work is generally in a relaxed fashion. Additionally,

pressure builds only over time and peaks when one’s funding ﬁnishes up. Often with graduate students, there

are additional psychological factors that build along with this.

Industry in contrast has a completely diﬀerent culture and work pace. There the work pressure is generally

higher than in academia but ﬂat over time. Although the monetary returns are way higher than in academia, the

1.2 Agile Research

quality may not be as high as an academic result. The work methodology matters a lot in the industry and that

discipline is what makes it thrive and produce tangible output. Both time and money are the driving factors for

constantly generating output.

Bringing the planning methodology and discipline of industrial work can have a huge impact on the work

style and throughput of academic work.

1.2 Agile Research

In industry, every development is conceived as a “Product”. All projects are designed to deliver a product

at the end of its execution. There is a deﬁnition of a product that can be applied across diﬀerent industries.

Product is deﬁned as the result of a process that provides a basic utility and is functional to serve a purpose.

This essentially means every solution ends up as a product. The key features of a product are the following.

A product does not have fancy capabilities but features.

Features may be added later but are not a basic requirement.

Product deﬁnition does not change during the process.

These key features also lay some foundations on how to develop a product. Designing and delivering

a product in the industrial setup follows a strict and disciplined method. Even though there are multiple

methodologies, one of the most popular and recent is called the ”Agile” methodology. The following section

explains the formal framework for how a product is designed in an agile methodology.

1.2.1 Problem Deﬁnition

The ﬁrst step is to deﬁne the problem. This involves clear articulation of the ﬁnal solution and the means to

attain that objective. There must be utmost clarity about the ﬁnal form of the solution and similar to a storyline,

there must be a well-deﬁned starting and a ﬁnal acceptance condition. Without this clarity, the next steps cannot

be deﬁned. The problem is deﬁned in a very strict manner such that the deﬁnition itself will generate the rest of

the requirements.

1.2.2 Scope

A solution for a problem can have diﬀerent levels of scope. There might be a solution that is very future-

proof and one that can cater to a limited set of use cases. The scope of the product/solution has to be deﬁned

along with the solution deﬁnition. This will determine the eﬀort required to build the solution. The scope

also helps to put speciﬁc boundaries around the solution which dictate its initial abilities. It will contain the

outcomes which will ﬁt a deﬁnitive timeline. The initial scope may have limited abilities, but this can iterated

with improvements in the next version.

1.2.3 Requirements

Once the problem is deﬁned along with its solution with a speciﬁc scope, the next item it will automatically

dictate is the set of requirements to build the solution. It is similar to saying for a speciﬁc dish, the recipe

requires the following ingredients. The scope will automatically dictate the minimum necessities for the diﬀerent

stages of implementation. When working as a team, the requirements will also specify “what” is required from

“whom”. Since the scope of the solution is already set, the requirement will not and should not change during

the execution of the project. It is also important to note that the requirements cover all aspects of the solution.

1.3 Process of Agile Execution

1.2.4 Timeline

The scope and requirements already give direction for the project execution. The next piece is deﬁning the

timeline. Often in industry, the timeline is decided by the budget of the project with other commercial reasons.

In academia, the duration of the funding and the deadline one might need to submit a thesis or report can be the

guiding principle.

In agile methodology, the whole solution is divided into independent individual parts before execution.

The independence of each part allows for setting speciﬁc milestones for the project and also testing along the

execution process. Figure 2 shows a representation of how the timeline is split across milestones. When the

Figure 2: Project execution is broken into diﬀerent modules with milestones across the time period of execution.

timeline is designed, it always made sure, that it considers the following aspects.

Know-how of the team members about the subject.

Availability of the team members.

Upcoming leaves and holidays in the time period.

Immediate availability of execution resources such as computational power and data.

The project lead also makes sure that the timeline also caters for any unexpected events that may happen during

the course of the project. To cater for such situations a minimum of 35% buﬀer is applied to all timeline

calculations. The timeline is initially set in days assuming that the person works for 9hrs a day. It will also

include research and study time if a critical piece of technology is required for the solution is new to the team

members.

The general principle of designing the timeline is to be extremely realistic about the execution of a task.

The time required should not be underestimated or even overestimated. It must be the most realistic value with

some buﬀer.

1.2.5 Risk

Every project/ solution has some risks associated with it. This may involve risks based on cost, the

relevance of the solution at the time of delivery, certain pieces of technology becoming obsolete, or even the

funding source even stopping. Such risks are agreed upon by all the team members and also the receiver of the

solution. The project always starts with agreement of such risks and countermeasures to tackle them.

1.3 Process of Agile Execution

In the agile methodology, it is the process of project execution that makes sure that intermediate milestones

are met and the solution is delivered on time with the required quality. The key highlight of the agile methodology

is its cyclic process which guarantees a reﬁned output and is never blocked during execution. Figure 3 shows a

representation of the agile process.

A general trend across all projects is that the initial solution that was envisioned during the problem

deﬁnition and scope design may not always work. Or it may have unforeseen deﬁciencies. In a traditional

1.3 Process of Agile Execution

Figure 3: The diﬀerent stages of an agile development methodology.

execution method, such scenarios may even stall the further steps of the project. Agile methodology is generally

immune to such risky situations. For each individual module to achieve a milestone detailed in the timeline, that

module will undergo the process of design, development, and testing. Once the testing phase of that speciﬁc

individual module is reached, it will decide whether at that point the milestone can be achieved or not. If the

testing fails, the milestone is redesigned with appropriate corrective measures, developed, and again tested. This

is a rapid process and can cater to both improvements and course corrections along the way. To show a broader

picture of the agile advantage, see Figure 4.

Figure 4: Comparison of waterfall vs agile method for project execution.[Image Credit: Google-Images]

The traditional approach to development is called the waterfall method. As the name suggests, the outcome

is only visible at the end of the timeline. The biggest ﬂaw of this method is that there is no intermediate

testing and no room for course correction along the timeline. In contrast, the agile method which breaks the

whole solution into individual independent components makes it possible to intermediate tests and also course

correction. This guarantees product quality and timely delivery of the solution. It is also noted that each

1.4 Reward function

intermediate milestone period is referred to as a “sprint”. Every sprint makes sure there is an output that can be

measured.

1.4 Reward function

No work methodology is successful without the healthy mental state of the team members. The team

members should feel a sense of accomplishment and work satisfaction in the process. A sense of achievement at

the end of the development cycle will generate plenty of energy to work on future projects. Work methodologies

have to make sure the team members of a project get professionally rewarded when they go through the process

of development.

Consider the traditional waterfall methodology with developing a car as shown in Figure 5. Let’s say the

development is spread over 4 steps. Step 1 makes a wheel and progresses to step 2 by joining them. In the third

Figure 5: Development of a car with waterfall methodology.[Image Credit: Google-Images]

step, the engine parts and some body parts are ﬁxed. With step 4 the car is completed. The most evident part

of this development life cycle is that the team members will only get a sense of satisfaction at the end of the

development. With the waterfall mechanism we have seen there are potential risks of failure making the ﬁnal

reward uncertain.

This is where the agile methodology makes sure that every milestone increases the level of happiness by

small factors such that the team members need not require to wait till the end to get rewarded. Figure 6 illustrates

this. With agile methodology, if one needs to develop a car, one would start with the smallest proof of concept

and progress in an incremental fashion. So one ﬁrst makes a skateboard, which is still not a car but you can

still move. Then you progress making a bicycle and then a motorcycle and ﬁnally a car. The main advantage

is that there is a far better incremental sense of happiness that the team members get over time and this always

increases the work spirit.

This is a key driving factor in the case of academic work. When a graduate student works on a research

project, his spirits are high in the beginning which often taper down over time and then there is a ﬁnal spike

at the end. This is basically the eﬀects of waterfall methodology and always demotivates young minds. But if

the agile style is followed there is always an intermediate milestone that lifts the spirit of people and no loss in

enthusiasm over time.

1.5 Conclusion

This article mainly elaborated the beneﬁts of following an agile work style for improving eﬃciency as well

as generating tangible outcomes. Mirroring these techniques may not be as straight forward in academia but is

still doable to a large extend.

When it comes to academic research, the ﬁnal product is always a publication. One way to think about

articulating each milestone would be to have each section of the publication completed at each stage. So for

each section to have a content, a speciﬁc set of milestones in terms of work has to be achieved. This can be

1.5 Conclusion

Figure 6: Product execution reward function with waterfall and agile method.[Image Credit: Google-Images]

incrementally done and ﬁnally the publication is ready to be send over for submission. Therefore here the

product is always a paper and the implementation path is to ﬁll in each section with the actual work done. This

always gives room for retrospective and improvement. Most importantly it guarantees increase in happiness

over time with reasonable output.

In academic research there are often many unknown factors which is diﬃcult to measure. With agile

methodologies, there is no clear direction on how to handle such uncertainties. This is challenge with fully

implementing this method in academic research. But this is where general academic practices of working on

proof of concepts and then testing will help.

Overall there is good beneﬁt in applying agile methods in research which improves eﬃciency, happiness

and also throughput of academic work.

About the Author

Dr.Arun Aniyan is leading the R&D for Artiﬁcial intelligence at DeepAlert Ltd,UK. He comes from

an academic background and has experience in designing machine learning products for diﬀerent domains. His

major interest is knowledge representation and computer vision.

The Ten Years of Science

by Ninan Sajeeth Philip

airis4D, Vol.1, No.5, 2023

www.airis4d.com

2.1 Development across sciences

Signiﬁcant scientiﬁc developments have occurred across various ﬁelds in the past ten years. From astronomy

to microbiomes, discoveries and technological advancements have opened up new avenues for research and

exploration. We have witnessed a remarkable explosion of scientiﬁc advancements that have transformed our

understanding of the world and revolutionised our lives. These developments have expanded our knowledge and

improved the quality of life with medical, energy, and communications breakthroughs.

Astronomy is one of the most exciting areas of scientiﬁc progress in the past decade. With new telescopes,

probes, and observational techniques, astronomers have discovered many new planets, stars, and galaxies. From

the detection of gravitational waves to the discovery of the ﬁrst interstellar object, the past decade has been a

golden age for space exploration, oﬀering us new insights into the origins and nature of the universe.

But the advances in science have not been limited to the cosmos. The past decade has also seen tremen-

dous progress in genetics and biotechnology, with the development of CRISPR-Cas9 technology, which has

revolutionised gene editing and made gene therapy a reality. The ability to manipulate genes has opened up new

possibilities for treating genetic disorders and developing treatments.

Another area of scientiﬁc progress in the past decade has been studying microbiomes, the complex

communities of microorganisms that live on and within us. Research has shown that these microbiomes play

a critical role in our health and well-being, inﬂuencing everything from our immune system to our mental

health. Advances in microbiome research have opened up new avenues for developing personalised medicine

and treating a range of diseases.

The past decade has also seen tremendous progress in neuroscience, with new insights into the workings

of the human brain and the development of brain-computer interfaces that can potentially transform the lives

of people with severe disabilities. With these advances in mind, the future of science and technology is bright,

with the potential for continued breakthroughs in medicine, energy, and communications that will transform our

lives and our world.

This article series will overview some of the most signiﬁcant scientiﬁc developments in the past decade.

2.2 Astronomy Research

Astronomy has witnessed unprecedented progress in the past decade, with discoveries and technological

advancements that have opened up new avenues for research and exploration. These developments have

2.3 Gravitational Wave Detection

signiﬁcantly expanded our understanding of the universe and have provided us with new insights into the nature

of the cosmos.

2.3 Gravitational Wave Detection

One of the most signiﬁcant astronomical developments in the past decade was the detection of gravitational

waves, predicted by Albert Einstein’s general theory of relativity over a century ago. According to the theory,

when massive black holes and neutron stars merge, they can produce ripples in space-time that propagate across

the universe. If we have a sensitive interferometer, such ripples in space-time can be observed as fringes

appearing in a speciﬁc pattern. The details about the colliding bodies can then be inferred from the nature of

the observed fringe.

2.3.1 Challenge

The interferometer was known even before Einstein’s General relativity was formulated. In fact, the famous

Michelson–Morley experiment in 1887 inspired Einstein to develop the theory of relativity. The Michelson

interferometer looks like what is shown in Figure 1.

Figure 1: The schematic of the Michelson Interferometer.

It has a laser beam of coherent light source that is split into two components using a semitransparent mirror

so that one part of it goes to mirror M1 and the other to mirror M2. These portions of the interferometer are

called the arms of the interferometer; the farther they are, the better the interferometer’s sensitivity. When they

are reﬂected back, they both are collected by the detector. You might have observed colour fridges when a thin

ﬁlm of oil is reﬂected in sunlight over clear water. The phenomenon is called interference, where depending

on the wavelength of the light reﬂected oﬀ from the two surfaces of the oil ﬁlm and its thickness, the waves

may cancel or enhance each other. Since sunlight has seven colours and the oil ﬁlm has varying thicknesses

depending on how it spreads over the water surface, diﬀerent wavelengths get cancelled and enhanced at each

point. This gives the ﬁlm such a beautiful colour pattern. But a laser beam is monochromatic, meaning it has

2.3 Gravitational Wave Detection

only one wavelength. That means, when they interfere, the fringes that are formed will be black and white.

Black where they cancel and white where they enhance. Now, if the arms are of the same length and the

mirrors are exactly perpendicular to each other, The light waves reﬂected from their centres will be identical in

movement. 1 We say they are in the same phase. So they will enhance each other and create a white spot at

the centre of the detector. But oﬀ-centre points will be farther for the beam of light, and hence there will be

a phase diﬀerence between their paths when they reach the detector. We will get a dark spot in regions where

the phases are opposite. Since these points are symmetric about the centre, what one observes will be circular

black and white ring structures caused by the subsequent in phase and out of phase light waves.

Since gravitational waves are very feeble, they might often be hidden in the random noise, making in

extremely diﬃcult to detect. To detect such feeble signals, our interferometer have to be super sensitive. That

means, they should have very sensitive mirrors and should have suﬃcient arm lengths to produce interference

patterns. In fact, the required arm lengths to detect even super massive blackhole merges would be a few

kilometers. Making such an interferometer is a technology challenge even today. It is so sensitive that even a

person walking at some distance from the interferrometer can create patterns! How will we know which one is

real gravitational wave?

The theory of relativity tells us the nature of the gravitational wave ripples produced during each type

of mergers. If we convert them inot sound waves, it will appear like a chirp sound. I would recommend

readers to listen to the chirp sound produced during blackhole mergers here: https://www.ligo.caltech.edu/

video/ligo20160615v2 The actual pattern would look as shown in Figure 2.

Figure 2: LIGO measurement of the gravitational waves at the Livingston (right) and Hanford (left) detectors,

compared with the theoretical predicted values Image curtesy :https://en.wikipedia.org/wiki/First observation

of gravitational waves

That is a clue, but because the gravitational waves might be completely hidden in the ambient noise, it is

almost impossible to detect them. Scientists developed what is called a matched ﬁlter that can ﬁlter out and

1In reality, there is a phase diﬀerence between reﬂected and refracted light. But for simplicity, we ignore it here.

2.4 Exoplanets

enhance probable gravitational wave signals from noisy signals they receive at the interferrometer. To make sure

that they are actually seeing a gravitational wave, they created two interferometers at two diﬀerent locations. A

real gravitational wave should make simultaneous detection of the patterns in both detector. Surely, the more

the number of detectors, the better would be the reliability of the detection.

In 2015, scientists at the Laser Interferometer Gravitational-Wave Observatory (LIGO) announced the

ﬁrst-ever detection of gravitational waves resulting from two colliding black holes 1.3 billion light-years away.

This discovery conﬁrmed the existence of gravitational waves and marked a signiﬁcant milestone in studying

the universe.

The success of the detection was not just the success of the General Relativity. It was also a demonstration

of the level of measurement precision modern instruments can oﬀer. The amount of ripple that the chirp signal

created on the LIGO mirrors were as small as the change in the distance to our nearest star that is 4 light years

away by a factor of the width of a hair! Even Einstein never expecte that we would be able to attain such a

measurement precision.

2.4 Exoplanets

The past decade has also seen signiﬁcant progress in understanding the formation and evolution of planets

and planetary systems. The discovery of exoplanets - planets orbiting stars outside our solar system - has

exploded in the past decade. When a planet moves around a star, if it happens to be in the line of sight of the

observer, it would cause a dip in the amount of light observed. A planet will produce a box type dip as shown

in Figure 3. The method is called transit method. See https://exoplanets.nasa.gov/faq/31/whats-a-transit/ for an

illustration. Following a transit method, the Kepler space telescope discovered thousands of exoplanets, which

detects the slight dip in light as a planet passes in front of its host star. The TESS space telescope, launched in

2018, has continued this work and discovered numerous exoplanets in its ﬁrst few years of operation. The James

Webb Space Telescope has just started revolutionising our ability to study exoplanets and their atmospheres,

giving us insight into the potential for life beyond our solar system.

Figure 3: The schematic of the transit method for exoplanet discovery. Image courtesy:https://astronomy.com/

siteﬁles/resources/image.aspx?item=%7BFAAA90E5-5808-4275-9EA8-3AD601DD31A4%7D

2.4 Exoplanets

In 2017, the discovery of the ﬁrst interstellar object, ’Oumuamua, provided astronomers with a rare glimpse

into the origins of our solar system. This cigar-shaped object was the ﬁrst known interstellar visitor to our solar

system passing through the inner solar system. Its unusual shape and behaviour suggest that it originated from

outside our solar system and emerged from another star system.

Another signiﬁcant development in the past decade has been the launch of new telescopes and observatories.

The Atacama Large Millimeter Array (ALMA), a radio telescope located in Chile, has opened a unique window

into the universe, allowing astronomers to observe the cosmos at longer wavelengths than ever. The Event

Horizon Telescope (EHT) has produced the ﬁrst-ever image of a black hole at the centre of the Messier 87

galaxy using a global network of telescopes.

Finally, the past decade has also seen signiﬁcant progress in studying dark matter and dark energy, the

mysterious substances that comprise most of the universe. The Large Hadron Collider (LHC) in Switzerland

has conducted experiments to search for dark matter particles. At the same time, the European Space Agency’s

Euclid will be launched on a Falcon 9 in July 2023. Following a travel time of 30 days, it will be stabilised to

travel a Lissajous path of large amplitude (about 1 million kilometres) around the Sun-Earth Lagrangian point

L2. It will study dark energy and dark matter through observations of the universe’s structure and evolution.

All these developments have made the past decade a golden eara for astronomy research. The story is

becoming even more exciting with new telescopes of bigger size are being built and commissioned. However,

the universe still remains a mystery to mankind and astronomy is still in its infancy.

[Will continue in next issue]

About the Author

Professor Ninan Sajeeth Philip is a Visiting Professor at the Inter-University Centre for Astronomy

and Astrophysics (IUCAA), Pune. He is also an Adjunct Professor of AI in Applied Medical Sciences [BCMCH,

Thiruvalla] and a Senior Advisor for the Pune Knowledge Cluster (PKC). He is the Dean and Director of airis4D

and has a teaching experience of 33+ years in Physics. His area of specialisation is AI and ML.

About airis4D

Artiﬁcial Intelligence Research and Intelligent Systems (airis4D) is an AI and Bio-sciences Research Centre.

The Centre aims to create new knowledge in the ﬁeld of Space Science, Astronomy, Robotics, Agri Science,

Industry, and Biodiversity to bring Progress and Plenitude to the People and the Planet.

Vision

Humanity is in the 4th Industrial Revolution era, which operates on a cyber-physical production system. Cutting-

edge research and development in science and technology to create new knowledge and skills become the key to

the new world economy. Most of the resources for this goal can be harnessed by integrating biological systems

with intelligent computing systems oﬀered by AI. The future survival of humans, animals, and the ecosystem

depends on how eﬃciently the realities and resources are responsibly used for abundance and wellness. Artiﬁcial

intelligence Research and Intelligent Systems pursue this vision and look for the best actions that ensure an

abundant environment and ecosystem for the planet and the people.

Mission Statement

The 4D in airis4D represents the mission to Dream, Design, Develop, and Deploy Knowledge with the ﬁre of

commitment and dedication towards humanity and the ecosystem.

Dream

To promote the unlimited human potential to dream the impossible.

Design

To nurture the human capacity to articulate a dream and logically realise it.

Develop

To assist the talents to materialise a design into a product, a service, a knowledge that beneﬁts the community

and the planet.

Deploy

To realise and educate humanity that a knowledge that is not deployed makes no diﬀerence by its absence.

Campus

Situated in a lush green village campus in Thelliyoor, Kerala, India, airis4D was established under the auspicious

of SEED Foundation (Susthiratha, Environment, Education Development Foundation) a not-for-proﬁt company

for promoting Education, Research. Engineering, Biology, Development, etc.

The whole campus is powered by Solar power and has a rain harvesting facility to provide suﬃcient water supply

for up to three months of drought. The computing facility in the campus is accessible from anywhere through a

dedicated optical ﬁbre internet connectivity 24×7.

There is a freshwater stream that originates from the nearby hills and ﬂows through the middle of the campus.

The campus is a noted habitat for the biodiversity of tropical Fauna and Flora. airis4D carry out periodic and

systematic water quality and species diversity surveys in the region to ensure its richness. It is our pride that

the site has consistently been environment-friendly and rich in biodiversity. airis4D is also growing fruit plants

that can feed birds and provide water bodies to survive the drought.