Courses Which of these Machine Learning courses should I attend?

AI Thread Summary
The discussion centers on selecting machine learning courses for someone studying Physics with programming experience in Java, MATLAB, and Python. Course 1 provides foundational knowledge in machine learning concepts and algorithms, making it a necessary prerequisite before tackling Course 2, which focuses on practical applications in engineering and science. Course 3, centered on TensorFlow, is considered beneficial but can be pursued later, especially if a relevant project arises. Participants emphasize the importance of understanding both theoretical and practical aspects of machine learning to effectively apply it in physics. Overall, a sequential approach of taking Courses 1 and 2 first is recommended, followed by Course 3 as time permits.
Wrichik Basu
Science Advisor
Insights Author
Gold Member
Messages
2,180
Reaction score
2,717
I am currently pursuing my undergraduate studies in Physics. While I do know some programming (Java, MATLAB, Python), I have no knowledge in Machine Learning. This is an important field, and I want to study it. I have gone through some online courses, and need your help in determining which of these I should attend.

Just to make things clear, I am doing these courses in my own time, and will not get any grades/certificates. These are just to enrich my knowledge (and also gain some experience, if possible).



Course 1: Introduction to ML

Course outline:
This course provides a concise introduction to the fundamental concepts in machine learning and popular machine learning algorithms. We will cover the standard and most popular supervised learning algorithms including linear regression, logistic regression, decision trees, k-nearest neighbour, an introduction to Bayesian learning and the naïve Bayes algorithm, support vector machines and kernels and neural networks with an introduction to Deep Learning. We will also cover the basic clustering algorithms. Feature reduction methods will also be discussed. We will introduce the basics of computational learning theory. In the course we will discuss various issues related to the application of machine learning algorithms. We will discuss hypothesis space, overfitting, bias and variance, tradeoffs between representational power and learnability, evaluation strategies and cross-validation. The course will be accompanied by hands-on problem solving with programming in Python and some tutorial sessions.

Course plan:

1642953147718.png




Course 2: Machine Learning for Engineering and Science Applications

Course Outline:
Functional programming is an elegant, concise and powerful programming paradigm. This style encourages breaking up
programming tasks into logical units that can be easily translated into provably correct code. Haskell brings together the
best features of functional programming and is increasingly being used in the industry, both for building rapid
prototypes and for actual deployment.

Course plan:

1642953483882.png




Course 3: Practical Machine Learning with Tensorflow

Course Outline:
This will be an applied Machine Learning Course jointly offered by Google and IIT Madras. We will cover the basics of Tensorflow and Machine Learning in the initial sessions and advanced topics in the latter part. After this course, the students will be able to build ML models using Tensorflow.

Course Plan:

1642953959420.png




My thoughts:

Course 2 is definitely geared for students in Science, so I want to attend that. However, it seems that it does not give a proper basic understanding of ML, which is covered by course 1. Course 2 has more of practical applications of ML, so I believe I have to do both, but course 1 followed by 2.

I am not sure about the third one, though. It focuses on TensorFlow. Should I do this later, if I find time? Will this be useful in the future if I pursue higher studies in Physics?

Looking forward to your advice.
 
Physics news on Phys.org
Once you take the second course then I imagine you could jump to reading some books like:
- 100 page ML book by Burkov
- Hands-On ML using Scikit-Learn Learn, Keras, and Tensorflow by Geron

Burkov’s book is freely available online as he allowed it to be shared and if liked bought.

Geron's book gives you a soup to nuts with a lot of practical knowledge in collecting, using, and testing your ML models.
 
  • Informative
Likes berkeman and Wrichik Basu
jedishrfu said:
Once you take the second course then I imagine you could jump to reading some books
Thanks for the books. I am more inclined towards practical knowledge so that I can use it in physics when necessary. Will definitely follow the books you mentioned.

What is your take on the first course? Does it seem like a necessary prerequisite for the second one? Or can I directly jump to the second one?
 
it looks good and it can’t hurt to take it. Basically, you need answers to these questions:
- how to identify what method works best for a given ML problem?
- what are the pros and cons, strengths and weaknesses of a given method?
- how to prep the data and select the training data and test data?
- how to use the method effectively?
- how to diagnose what is going on when they fail?

Many of these methods are essentially black boxes where you are taught the theory of how they work but not necessarily how they are actually implemented. This means that they may fail in odd ways under rare circumstances and it would be hard to figure out what went wrong.
 
  • Informative
Likes Wrichik Basu
Wrichik Basu said:
I am currently pursuing my undergraduate studies in Physics. While I do know some programming (Java, MATLAB, Python), I have no knowledge in Machine Learning. This is an important field, and I want to study it. I have gone through some online courses, and need your help in determining which of these I should attend.

Just to make things clear, I am doing these courses in my own time, and will not get any grades/certificates. These are just to enrich my knowledge (and also gain some experience, if possible).



Course 1: Introduction to ML

Course outline:Course plan:

View attachment 295891



Course 2: Machine Learning for Engineering and Science Applications

Course Outline:Course plan:

View attachment 295892



Course 3: Practical Machine Learning with Tensorflow

Course Outline:Course Plan:

View attachment 295894



My thoughts:

Course 2 is definitely geared for students in Science, so I want to attend that. However, it seems that it does not give a proper basic understanding of ML, which is covered by course 1. Course 2 has more of practical applications of ML, so I believe I have to do both, but course 1 followed by 2.

I am not sure about the third one, though. It focuses on TensorFlow. Should I do this later, if I find time? Will this be useful in the future if I pursue higher studies in Physics?

Looking forward to your advice.

You can take all 3. They don't overlap significantly. 1 and 2 both cover some important fundamentals. I would start with 1, then 2. 3 would be good to take in combination with a serious project if you have one in mind.
 
  • Like
Likes Wrichik Basu and jedishrfu
I'm going to make this one quick since I have little time. Background: Throughout my life I have always done good in Math. I almost always received 90%+, and received easily upwards of 95% when I took normal-level HS Math courses. When I took Grade 9 "De-Streamed" Math (All students must take "De-Streamed" in Canada), I initially had 98% until I got very sick and my mark had dropped to 95%. The Physics teachers and Math teachers talked about me as if I were some sort of genius. Then, an...
Bit Britain-specific but I was wondering, what's the best path to take for A-Levels out of the following (I know Y10 seems a bit early to be thinking about A-levels, but my choice will impact what I do this year/ in y11) I (almost) definitely want to do physics at University - so keep that in mind... The subjects that I'm almost definitely going to take are Maths, Further Maths and Physics, and I'm taking a fast track programme which means that I'll be taking AS computer science at the end...
After a year of thought, I decided to adjust my ratio for applying the US/EU(+UK) schools. I mostly focused on the US schools before, but things are getting complex and I found out that Europe is also a good place to study. I found some institutes that have professors with similar interests. But gaining the information is much harder than US schools (like you have to contact professors in advance etc). For your information, I have B.S. in engineering (low GPA: 3.2/4.0) in Asia - one SCI...
Back
Top