Machine Learning 101: How & Where To Start For Absolute Beginners.
This post covers all you will need for your Journey as a Beginner. All the Resources are provided with links. You just need Time and Your dedication.
I have separated this post into several programs as shown below:
Path A. 4 to 5 Months
Path B. 2 Months or Less
Path C. 1 Month or Less
Path A: Learn In 4 to 5 Months.
Part 1: Start With Machine Learning, 2 months.
- Start Coursera Machine Learning course by Stanford University (coursera.org/learn/machine-learning)
- By the time you start Neural Network on week 5 at the Coursera course, complete the Neural Networks playlist by 3BIue1Brown.
- I think the Coursera course rushes the Neural Network part a little bit, there is a fantastic free online book on Neural Network and Deep Learning (neuralnetworkanddeeplearning.com)
- Now many of the concepts in Machine Learning and Deep Learning literature will start making sense to you. For fun, head over to Chriss Olah’s blogs, they’re awesome! (http://colah.github.io/)
Part 2: Deep Learning, 1 month.
-Before you start Deep Learning, you need to brush up some university math. The Deep Learning Book by Ian Goodfellow, I would recommend you go through Linear Algebra and Probability and Information Theory chapters as deeply as you can.
- My best pick to start Deep Learning is with Andrew Ng’s Deep Learning specialization. (coursera.org/specializations/deep-learning)
- It’s also time to read Chapter 3, 4, 5, 6 from neuralnetworkanddeeplearning.com to strengthen your concepts
Timing: if you are working full time on these courses, l think it’s possible to finish each week’s content in 1~2 days. So don’t get intimidated by
the schedule. But give yourself some time to breathe between courses.
Part 3: Practical Implementation of Deep Learning (1~2 months).
- Fast.ai has a wonderful resource for practical Deep Learning (course.fast.ai), While Andrew Ng or others teach in a Top-down approach (know first, do later) fast.ai teaches in a bottom-up
approach (do first, know later). - Two other courses I would mention is CS231n and CS224n by Stanford University. CS231n is focused on computer vision with Deep Learning,
and CS224n focuses on Sequence Modeling such as Natural Language Processing with Deep Learning.
Path B: Learn in 2 Months or Less.
Complete the first 5 weeks of the Machine Learning course from Coursera, Do the programming exercises.
- Watch the Neural Network playlist from 3Blue1Brown youtube channel.
- Complete Course (Neural Networks and Deep Learning) from Deep Learning Specialization in Coursera. Do the exercises.
- if you want to start an Image Processing project, take the 4th-course ln Coursera specialization or if you want to work on Natural Language Processing or sequence data, take course no.5.
- Search for open source implementation and YouTube videos of projects that you are interested in. if you are concerned about which language to use, I think it’s good to stay with Keras (Keras is an open-source neural network library written in Python)
Path C: Learn in 1 Month or Less.
-Skim through Coursera Machine Learning course Week 1 to 5. Just watch the videos, grasp the concept. You can skip the MATLAB/Octave tutorials in Week 3.
- Watch the Neural Network playlist from 3Blue1Brown youtube channel.
- Skim through Course (Neural Networks and Deep Learning) from Deep Learning Specialization in Coursera.
- If you want to do an Image Processing project read the chapters from Nielsen’s book: neuralnetworksanddeeplearning.com/chap6.html
- Siraj Raval has some interesting videos to give you a gist of most ML and DL topics.
- Search for open source implementation and YouTube videos of projects that you are interested in. And keep tweaking them to your need
Links & Optional Resources.
Neural Networks and Deep Learning:
coursera.org/learn/neural-networks-deep-learning
Siraj Raval: Youtube Channel
- I would suggest you to follow 2 minutes Paper on YouTube to get
updated with the wonders that researchers are doing with Deep Leaning around the world.
