Sunday, 23 May 2021

Second Semester Thoughts | M.Tech. (Data Science & Engineering) | BITS Pilani WILP

 Hi everyone, 


This is another post regarding the M.Tech. in Data Science & Engineering program.

If you haven't read my earlier post - please do so here - https://rupakh.blogspot.com/2020/04/mtech-data-science-engineering-wilp.html



Important Disclaimer: Please note that this post has to be taken only as advice and not a formal directive from the university. I am not paid to write this post nor do I take any responsibility that arises if you decide to make a decision based on my post.

In this post, I will be discussing my experience having gone through (and passed!) the second semester here.

It will help if I go subject by subject so let me take this approach. 


Introduction to Statistical Methods:

This is a very important course with respect to understanding the basics of statistics. The course begins with an introduction to probability which has very important implications and applications in machine learning (probabilistic learning - Chapter 6 of Tom Mitchell  - Bayesian Learning) and in artificial intelligence (Hidden Markov Models) and in probabilistic graphical models in Semester 3. We then move on to mean, median mode, standard deviation, and variance concepts which will help you understand the core of statistics. Then, we will study statistical distributions, and move on to Hypothesis Testing. Each and every chapter is linked to the previous chapters so do not miss even a single class. We then study regression (which helps you understand regression studied in ML much better) and then finally to Time Series Analysis. 

Time Series Analysis will be a separate topic and it is very interesting. 

The textbooks for this course are:

Probability and Statistics for Engineering and Sciences,8th Edition, Jay L Devore, Cengage Learning (https://www.amazon.in/Probability-Statistics-Engineering-Sciences-9E/dp/9353506247)

and 


These two textbooks will suffice and it will be better to have the hard copies of these textbooks even for future purposes. 

Our examinations was online and 3 out of 6 problems were to solve in in Microsoft Excel spreadsheets for this subject especially for Time Series Forecasting module. So, make sure to learn the theory well and also how to use Excel to implement the problems in the exam. 



Introduction to Data Science:

This is a theory subject with some overlap with Data Mining subject. 

I had used this book - Introducing Data Science:



This a very theoretical course in my experience, class lectures and notes were enough along with the textbook mentioned above. Nothing more to write about this subject except that it can be managed. 

Machine Learning:

This course is very important in this semester and will also help you in your future job search. We will learn basics of all machine learning techniques beginning with Bayesian Learning, Regression, Decision Trees (repeated from Data Mining), Neural Network (which will form the basis of Deep Learning in Sem 3), Instance-based Learning, Ensemble Learning, Support Vector Machine (very important concept), and finally Unsupervised Learning. We used to refer to lectures of Dr. Y V K Kumar. His lectures were the best!

Textbooks:

Machine Learning - Tom M. Mitchell (the best book for ML) - a must have!


This book should be more than enough along with the PPTs shared by BITS and the lectures. 

Artificial Intelligence:

I had opted for this elective since I was not interested in a hardware based course of Systems for Data Analytics -SDA (an extension of COA) and in data visualization course of Data Visualization & Interpretation - DVI.

I found this course quite difficult because, in my opinion, the faculty was not too great. The question papers were quite tough but NPTEL courses helped along with the standard textbook. The textbook needs to be read properly at least once and you will get a fair idea about the course. It is very well explained. The textbook is simply the bible for this course. No other books are needed.

NPTEL Courses: 



Artificial Intelligence - Russell/Norvig


Final word of advice:

Concentrate more on ACI and ML since these are AI is a 5 credit course and ML is a 4 credit course and will count more towards your CGPA. That being said, ISM is also an important course to helps you understand the statistical aspect of algorithms.

Please make sure that you do NOT miss the mid semester and the comprehensive examinations. You may have to repeat the course or you will get some RRA grade which means you will have to spend more time to finish this course. It is better to complete it within the time frame - all you need is to manage your work-life balance effectively. 

Try to score more marks in your quizzes and assignments as they form 40% of the marks. 

I will keep updating this page as and when I get time. But I hope this helped you all. 

I would love to expand my professional network on LinkedIn. Please send me a connection request here: https://www.linkedin.com/in/rupakh/

Thanks!

Cheers, 
Rupak.


at 3 comments:

Tuesday, 12 January 2021

Artificial Intelligence | Semester 2 | M.Tech. (Data Science & Engineering) | BITS Pilani WILP





Important Disclaimer: Please note that this post has to be taken only as advice and not a formal directive from the university. I am not paid to write this post nor do I take any responsibility that arises if you decide to make a decision based on my post. This logo does not belong to me and it is the sole property of BITS Pilani. 

Hello everyone,

In this blog post, I will be talking about the course, Artificial and Computational Intelligence (ACI) in the M.Tech. in Data Science & Engineering program offered by BITS Pilani through the WILP mode. 

This is a course named 'Artificial and Computational Intelligence' and is offered in the second semester of the M.Tech. program as an elective. There were two others electives offered - System for Data Analytics (SDA) and Data Visualization and Interpretation (DVI).

I do not know much about the other electives but will try and touch about them below:

SDA - This is a extension of COSS course in the first semester. It can be seen as COSS for data analytics. The course probably has some sections for the hardware aspects of data analytics as well.

DVI - As the name suggests it has to do with visualization of data with heavy usage of Tableau software.

Now, coming back to ACI, let me list the material that you need.


Artificial Intelligence: A Modern Approach by Pearson - Third Edition by Russel & Norvig: This is the best book for this course. The material is lucid and comprehensive. I feel that this book is more than enough.

The links to buy these books are given below:

On Amazon: 




I bought this book from Amazon for Rs. 505. It was a used copy in good condition. A used book should be good. 

I also bought two other books mentioned below. Both were used copies.

Artificial Intelligence: A New Synthesis by Nilsson  - https://www.amazon.in/gp/product/8181471903/

ARTIFICIAL INTELLIGENCE Third Edition by Knight, Rich, and Nair - https://www.amazon.in/gp/product/0070087709/

According to me, these two books are not needed for the course but can be a good addition to your data science library at home.


This handout is for B.E. students but the M.Tech curriculum is quite similar. 

The prerequisites for this course are - knowledge of a programming language (preferably Python) and a solid knowledge of Data Structures and Algorithms. The DSAD course in the first semester helps in laying the groundwork. Revise the concepts of algorithm design and graph theory - especially BFS and DFS and you are good to go!

This course primarily deals with how agents perceive the world and what actions do they need to take. Say for example, you need to build a self driving car - what does the car need to do in order to be a 'good' agent? The car should safely take its passengers from point A to point B. Thus, to be able to build such a system, we need to study the concepts of artificial intelligence by exploring search spaces, game playing, local search algorithms, constraint satisfaction problems,, first order logic, reasoning over time and reinforcement learning.

Overall, this course is really interesting so far and I hope I am able to apply what I study here in the near future. 

I also referred to NPTEL videos given below:



I liked Sudeshna Sarkar's of IITKGP videos. I haven't seen the IITD videos.

Hope this helps!

Cheers, 
Rupak 



at No comments:

Monday, 16 November 2020

Beautiful evenings at the beach!

Hi there!

This lockdown has given me a fantastic opportunity to rediscover the things I love doing the most.

One of them, particularly, is walking. It's a habit I picked up from Amma (my mother) and I wish to continue pursuing this habit till the day I die. 

I am fortunate enough to live close to the beach - a mere 20 minute walk from my house and I reach the first public entrance of Juhu beach. From thereon, I walk till I reach Sun-n-Sand hotel. Until now, I thought it was a 3 km walk one way but when I checked on Google maps, I can see that it is a 4 km walk! Feels good to have walked a km more daily! :)



This is the path that I follow. 

I like to stop my walk at Sun-n-Sand because it is relatively less crowded and I can get a spectacular view of the sunset from this spot. This is what the sunset looks like:




This picture does not do justice to the amazing view that I witness - it is best left unsaid and it is something that you must experience when in Bombay! :) 

I sit down here for about half an hour and continue watching the stunning colours of sunset. 

As a side note, please bear with me on the following content. It somehow connects to what I have written so far. Read on.

I've always been interested in how the human brain functions. Isn't is fascinating? Billions of neurons in the brain, all working with each other to make us function the way we are today. It is of special interest to me because it is my field of studies as well. I am going to be venturing in the field of data science - particularly in the field of machine learning. 

I got interested in the field of machine learning in my final year of engineering where I used artificial neural networks to price European Call options. You can read more about options here.

After that I started working in data warehousing and I drifted away from this field only to be back 7 years later as part of my graduate studies (coursework in M.Tech.) and my interest. 

I am currently in my second semester of M.Tech, and I have courses in Artificial Intelligence and Machine Learning. I would like to work in the area of machine learning, in future, where I would want to help companies identify patterns or anomalies in data and help solve business problems. An example would be to identify cancerous tumors in the brain by detecting unusual growth in parts of the brain. This actually is a subdivision of machine learning called deep learning, which I will learn in the third semester which deals exclusively with the workings of artificial neural networks. 

Deep learning is learning how the human brain works and how we can model it in computers to help solve problems or identify patterns. Let's say that in childhood your parents told you how jam looks and in which glass bottle it is stored. Over the years, you may have associated a jam bottle to look like this - at least I have:



Now whenever I see this bottle, I know that it will contain mixed fruit jam. It is this "learning" that needs to be taught to the computer by modelling an "artificial" network of neurons just as they're present in our brain. Obviously, it is a very difficult task to model 86 billion neurons! Complexity time alone would be in multiples of exponential time - a very costly endeavour in terms of time. One can also imagine the magnanimous amount of cost with respect to space. We would need petabytes of space!

Anyway, enough of my geekiness.

I've been trying to meditate. My time in South Africa was quite scary and things got a little rough for me, mentally, after I was shown a gun. I think I may have slight PTSD. You can read about it here.
To calm myself down, I've come to learn that meditation does wonders. This video helped me a lot understand the power of meditation and this is something I am trying to incorporate in my daily life. You should too! It's a brilliant video by an expert in consciousness who lectures about the power of meditation. 

I feel so happy to meditate at the beach! The last few rays of the sun falling on my face as I close my eyes and listen to the waves breaking. It is almost as if the waves are following the Normal distribution!
The waves start to rise from the left and right sides and peak in the middle and then suddenly all come crashing down. This cycle then repeats. The funny part is, I can only visualize this with my eyes closed and I am getting this information through my ears in the form of sounds waves as my primary source of input. I can also feel the winds becoming stronger as I feel the breeze making me happy. It is as if I am reaching another state - I need to explore this more as I read more on meditation and perhaps go through that Stanford lecture again. 

But all of this disappears as I stop meditating! This is quite funny - nothing really has changed, has it? The wind continues to blow, the waves continue to crash but why does it feel different? I think it has to do with my concentration. 

This got me thinking. Why did it feel different with my eyes closed? My visual input was blocked as I closed my eyes. I only had my ears and my skin working for me (my nose, tongue, and eyes were of no use in this situation). Through my skin, I was able to feel the cool breeze and my ears helped me with hearing the waves and simulate the Normal distribution-ish curve. 

Personally, this was an interesting breakthrough for me. It got my thinking how bad multitasking really was. I had been reading and hearing about is but to experience it first hand is something else. The level of focus that I reached while meditating for 5 minutes was unmatched! I will definitely try to implement this while studying, reading a book, or listening to music. Focus on one thing 100%!

This will also be how I model my neural network in future - taking in only the needed inputs and focusing on getting the right results. My network will focus on getting only ONE output perfectly, nothing else will matter, it will also NOT multitask and try to accomplish what it intended to achieve in the first place. Each network will have only one specific function. What a great takeaway, isn't it?

After all, I will be modelling my artificial neural network based on the biological neural network, so it helps to gain such insights!

Anyway, coming back to my day, I then proceed to head back home on the same path before clicking a few pictures at the entrance of the beach while enjoying the stunning dusk colours! See it for yourself:



Again, it looks more stunning to see it for yourself! Maybe you should come for a walk with me to experience these vivid colours!

I am so previewed, fortunate, and lucky to have Juhu beach close to where I live.

I feel so happy to visit this place almost daily because it offers me an opportunity to play with dogs, see so many types of people enjoying the beach, and most importantly it is part of my ongoing fitness plan - the results are to be seen. Let's see how that pans out.

Until then, 

Cheers!

Thanks for reading. 

Rupak



at No comments:

Friday, 13 November 2020

First Semester Thoughts | M.Tech. (Data Science & Engineering) | BITS Pilani WILP

Hi everyone, 

This is my second post regarding the M.Tech in Data Science & Engineering program.

If you haven't read the first post - please do so here - https://rupakh.blogspot.com/2020/04/mtech-data-science-engineering-wilp.html



Important Disclaimer: Please note that this post has to be taken only as advice and not a formal directive from the university. I am not paid to write this post nor do I take any responsibility that arises if you decide to make a decision based on my post.

In this post, I will be discussing my experience having gone through (and passed!) the first semester here.

Let me begin by saying that this course is for serious aspirants. Agreed, the admissions were easy and you did not have to clear the GATE examination but that does not meant that the course is also easy.

Remember, it is conducted by one of the best engineering colleges in the country and they have reputation to maintain. Diving your time between work and studies is of utmost importance to do well in this course. The sooner you master this point, the better it is!

It will help if I go subject by subject so let me take this approach. 


Data Mining:

This is a 3 credits course and is definitely one of the more important courses in this semester. This subject forms the base for Data Science & Engineering because in this course you will learn how to use raw data and run algorithms over it like k-means clustering, association rule mining, outlier analysis, etc. Do not take this course lightly because to learning everything in one go is really difficult and you will regret not studying regularly. The textbook of Han Kamber is more than sufficient along with YouTube videos. Practice more problems than learning theory. 

Mathematical Foundations for Data Science:

I will not beat around the bush - this is the toughest subject I have studied till date. I got a panic attack while giving my mid semester exam and I fainted. I was literally this stressed out! But you need not be. Read on.

This is a graduate level course in linear algebra which teaches you vector and matrix algebra, systems of linear algebraic equations and their solutions; eigenvalues, eigenvectors and diagonalization of matrices; graphs and digraphs; trees, lists and their uses; partially ordered sets and lattices; Boolean algebras and Boolean expressions. 

The first half of the course was brutal. I went through the lectures multiple times to understand concepts and referred to MIT and IIT lectures. 

These are the lectures I saw:



I also bought Gilbert Strang's book. It was really helpful. You will find it here: https://www.amazon.in/Linear-Algebra-Applications-Gilbert-Strang/dp/8131501728/

My advise is to watch each lecture as soon as it is taught on Saturday/Sunday and 0.75x speed and write everything down in a notebook and try and understand matrices and vector spaces. This will help you a lot in future courses like Machine Learning and Artificial Intelligence. 

If you aren't understanding everything, it is fine. Do NOT panic. In this course, even the toppers will struggle (a little less than others though). Because of relative grading, you will be fine if you are sincerely studying, submitting all assignments and quizzes on time and studying hard. 

And keep calm. Everything will be fine! :D

Computer Organization and Software Systems: 

This course deals with the hardware and software aspects of computers related to data science in an indirect way. For CSE/IT students, it is a recap of COA and OS learnt during BE/BTech.

The first half of the course is COA/Hardware aspects and the next half of the course if OS/Software aspects.

In my opinion, if you pay attention in class and are regularly studying, you should be fine. The marking in this subject is quite strict - don't know why. I got the worst grade in this subject - a B grade :(

I feel the latter half is easier because I liked the OS part more. The most tricky part in the first half was the design of the processor but this is a great lecture which helped me a lot - https://www.youtube.com/watch?v=0B-y1RPDXjs

Data Structures and Algorithms Design:

This course is also very important because data science effectively deals with building fast and optimal algorithms to process data efficiently. Our faculty was simply great and I enjoyed this course a lot because I like programming in general. It can be a tough course if you do not understand the concepts so be sure to study the concepts well. 


Even if you can get the used copy, do buy it, it is a must book for any computer scientist. 

Just before exams, you can refer to Prof. Abdul Bari's lectures. They were of great help. You can find them here: https://www.youtube.com/channel/UCZCFT11CWBi3MHNlGf019nw

Final word of advice:

Concentrate more of COSS and DSAD since these are 5 credit courses and will count more towards your CGPA. That being said, DM and MFDS are also important course to help set the right base for future courses so do not neglect them.

Please make sure that you do NOT miss the mid semester and the comprehensive examinations. You may have to repeat the course or you will get some RRA grade which means you will have to spend more time to finish this course. It is better to complete it within the time frame - all you need is to manage your work-life balance effectively. 

Try to score more marks in your quizzes and assignments as they form 40% of the marks. 

I will keep updating this page as and when I get time. But I hope this helped you all. 

I would love to expand my professional network on LinkedIn. Please send me a connection request here: https://www.linkedin.com/in/rupakh/

Thanks!

Cheers, 
Rupak.

at 3 comments:
Labels: , , , , ,

Tuesday, 21 July 2020

~Amazon Prime Day 2020



Hello everyone,

I know it has been a while since I've blogged and that is because I was really busy with some other commitments.

But, there is some great news!

Amazon Prime Day is back for 2020!

If you know about it, you know how awesome these 2 days are. But if you don't, read on!

Amazon India has a program called Amazon Prime which is a must have for each and every shopper in India. I will explain the benefits of this membership in this blog post.

1. Cost of Amazon Prime Membership:

The yearly cost of this membership currently is Rs. 999.

They also have a monthly option which costs Rs. 129 for a month - which comes out to Rs. 1548 for a year. Thus, it obviously makes sense to pay for the yearly option of Rs. 999.

Now comes the most important question - is it worth it?

Is the Amazon Prime membership worth paying Rs. 1000 for a year?

Speaking from my personal experience, I would say it is 100% worth it.

Why should you believe me?

You should believe me because I have no stake in Amazon, and I am just a happy customer :)

What are the benefits of Amazon Prime Membership?

To list a few, from the Amazon India website,

1. Quick and Free Delivery:

You get free 1 day or 2 day delivery on most items which are marked as "Prime".

This is one of the best deals. I order almost everything from Amazon, be it groceries, meat, and electronics. Heck, I even pay all my bills using Amazon. So you can trust me on this.

The products having "Prime" label helps you identify good products which have fast delivery.

Will these deliveries be free? YES! Fast and free delivery!

 2. Prime Video:

 In this time period of lockdowns and the Covid-19 coronavirus, I'm sure you must be streaming some shows on Netflix.

Try Amazon Prime Video if you haven't! It has a collection of great shows and yes, The Office (US) is included! Bears, Beets, Battlestar Galactica!

This Prime Video streaming service is included in the Prime membership cost.

3. Prime Music:

 I've never used Spotify or YouTube music because the music catalogue offered in Prime Music is outstanding! I listen to a lot of genres - Bollywood, Thrash Metal, Classical Music, Classic Rock, and more and I've never had a problem finding any of the songs! :)

They also have deals with respect to Prime Reading and Gaming, but I haven't had a chance to try these services out.

So... how much have I saved?

Numbers do not lie, so here are my savings:



As you can see, I have shopped a lot, and in the process, saved a lot as well!

Now, to avail the benefits of Prime Day, it is imperative that you have an Amazon Prime membership because I have noticed that Prime members get early access to deals and get better discounts.

Now, how can you get the best out of Prime Day in 2020?

1. Remember the dates - 6th and 7th of August!

2. Make a list of items you have wanted since a long time. Waiting to upgrade your mobile phone? Want to purchase a new sofa set? Prime Day is the time to purchase these products.

3. Hope you have credit/debit cards which help you save more money on purchases. According to past trends, Amazon had offers for American Express/ICICI Bank/HDFC Bank/SBI credit card holders. You can easily get an extra Rs. 1000 or Rs. 2000 on your purchases. This time around, they have chosen HDFC Bank credit/debit card holders for an instant 10% discount. If you do not have a HDFC Bank credit or debit card, fret not, ask your friends!

4. You can buy/renew your Prime membership for a lower rate. I had paid Rs. 899 for the membership I currently have. They had a discount of Rs. 100 on Prime Day! I am hoping for this offer this time as well! Let's hope they have it :)

5. They will start releasing deals to look out for from the 23rd of July - be sure to keep an eye for these offers!

There you have it! Exciting offers, lots of savings, and all the products that you need.

I am ready for Amazon Prime Day 2020! Are you? :)

Cheers,
Rupak
at 1 comment:
Subscribe to: Posts (Atom)

Second Semester Thoughts | M.Tech. (Data Science & Engineering) | BITS Pilani WILP

  Hi everyone,  This is another post regarding the M.Tech. in Data Science & Engineering program. If you haven't read my earlier pos...