AI Roadmap for my 1-year Younger Self
This is my first blog, so I guess it's gonna be very Naive. I hope it's gonna be like Reinforcement Learning in which my skills will improve with the rewards associated with writing blog posts :)
In this article, I will design a roadmap for the 1 year younger myself who knew nothing about ML or Data Science. He was looking for a thing which he would be interested in. If this roadmap was available to me at that time, I hope it would have made my past 1-year learning journey somewhat smooth and efficient.
I am going to talk about courses that one should start with and in what order.
1. Basic Python
Python is very easy to learn and implement. Python is a must for ML. When it comes to AI programming languages, Python leads the pack with its unparalleled community support and pre-built libraries (like NumPy, Pandas, Pybrain, and SciPy) that help expedite AI development.
I found this playlist on Youtube to be good and sufficient. You should watch till Video 9, that will be sufficient. You can watch the remaining videos if you have time.
2. Machine Learning by Andrew Ng
It's an ML course offered by Stanford University on Coursera by Andrew Ng. It’s a wonderful course, and after 2–4 weeks, you will understand whether you are having an interest in ML or not. After this course only, I understood that I like this field.
One Problem with this course is that the programming assignments are in Octaves. So, it's difficult for anyone to learn octave for this. The solution for this problem is that you can get Unsolved Assignments for this course in Jupyter Notebook files at IITG.ai’s Github repo.
IITG.ai ‘s Github Repo for Unsolved Assignments in ipynb format
How to get Coursera courses for free?
Method 1: Auditing the Course
This way, You’ll be able to see most of the course materials for free, but you won’t be able to submit certain assignments or get grades for your work. Plus, You won't get a certificate.
Method 2: Applying for Financial Aid
In Step 4, You are required to answer some questions in detail. You can google for ideal answers. Ensure that answer for “How will taking this course help you achieve your career goals?” should not be copy-paste; you should write related to the course you are applying for. After applying, your financial aid will be approved in 15 days. After that, you will be able to submit assignments and get the certificate after completing the course.
The best way is to apply for financial aid when you start and then audit the course. This way, you will get access to the course material, and after aid approval, you can submit the assignments.
3. Deep Learning Specialisation
Its specialization in Deep Learning is offered by deeplearning.ai and is taught by Andrew Ng on Coursera. It contains 5 superb courses which cover Basic Neural Nets, CNNs, NLP, etc. It also contains many cool, guided projects.
4. Getting started with TensorFlow 2, i.e., KERAS
It's a very nice course on Coursera, and it covers all key concepts in Keras Framework. In DL specialization, most of the assignments are in Keras, but it does provide you enough knowledge about Keras. This course is short and very effective.
5. Learning PyTorch
It is always advised to use the PyTorch framework instead of Tensorflow or Keras because the PyTorch community is vast now.
At the initial level of learning, you will find that it's very complicated, but after some time, you will realize that it's very convenient to use, and you can get your code doubts cleared very easily as the community is vast.
I suggest Pytorch Tutorial by Python Engineer as it covers all the key things that you should know. If you did the above Keras course, then this one will be very smooth for you as the order of things is similar.
The below-mentioned link is the Github repo for the code used in each video of the whole playlist. The link to the playlist is in the readme section. But I suggest that going through the ipymb code files will be sufficient. If you have doubts about the code, you can go to a particular video and resolve it. The videos are very long, and it would be very time-consuming to go through each video.
6. Fast.ai Deep Learning Course
fast.ai is a high-level API built on PyTorch. Fast.ai also provides some free courses that are very helpful.
There are two parts in the DL course by fast.ai. The first part covers most of the implementation part, and 2nd course covers more of the theoretical domain. I found this very helpful as it is in PyTorch.
They use the fast.ai library in codes, but it's up to you that you want to learn this library. But it's okay not to use this and use only PyTorch code.
There is also the fastbook repo, which contains two types of code for each lesson. One type is the notebook used by Jeremy in the lessons with all the theory material, and the other type has notebooks with only code cell and theory scraped. You can use the 2nd type to do experiments and all.
Use Google Colab instead of Jupyter Notebook as running on your laptop will consume a lot of Data as well as it would be not that computationally good as the Colab GPUs.
Part 1: Practical Deep Learning for Coders
OpenCV
OpenCV python tutorial on youtube is a perfect playlist. There are no prerequisites if you know basic python. You can go through this simultaneously with the above courses.
CS229, CS 224N and CS231N
These are some courses by Stanford University and are available on Youtube. They are slightly advanced, and one can do based on interest and availability of time.
CS229 is somewhat similar to the ML course by Andrew NG mentioned above. It contains the mathematical part a lot.
CS224N is for NLP, and CS231N is for Computer Vision.
Summer Analytics Course
This website can give you Intermediate knowledge about Data Science. It also contains Data Scraping by Python, Resources to learn Python, Pandas, Numpy, and much more.
This website is absolutely free to access.
Some Important Resources Link
- Download Jupyter Workspace files in Coursera
- Configuring Google Colab Like A Pro
- DL with PyTorch
- DL Course
- Albumentations (an augmentation Library)
- Feature Visualisation
You can work on some Projects and Competitions on Kaggle simultaneously. One more important thing is that you should start blogging sooner.
It's an excellent article by Rachael Thomas on why you should blog.
I hope this article will help in your journey.
I guess that's it for my first blog. See you in the next one :)
