Some thoughts on choosing resources
Before you choose a resource to get started, it makes sense to spend some time thinking about your learning style. When do you notice you are most engaged and learn best? Is it when you are reading, listening, writing, drawing, or maybe doing the thing you are learning?
Starting out with a resource that suits your own personal way of learning can save you some time and frustration down the road.
Self-learners, do you feel stuck in your progress? Try reevaluating your learning style.
Recently I've switched to a blend of reading and note-taking with a focus on making concepts more visual.
ð ð ðž
Instead of coding along with a course, I learn first and then build.17:42 PM - 14 Jan 2021
I'm still discovering my best learning style. I thought I learned better from listening, because college was one of the most productive times of my life. But I have since found that it probably wasn't the act of hearing, but the physical and visual act of attending lectures, and even more importantly--taking notes. The more visually interesting the notes the better. So I've been trying to develop my note taking skills to show the hierarchical structure of concepts, adding little diagrams, graphs, and doodles wherever possible.
If you've been finding your progress is flagging lately, maybe (re)assess your learning style. Maybe you are finding it hard to push through a course. You could try a book instead. Maybe books aren't for you. Try a tutorial where you read a little, code a little.
Maybe you're not sure what your learning style is yet! That's okay too! Just pick a resource at random and get started. It doesn't have to be the perfect one.
To make your learning more active, explore note-taking strategies. Do you learn best with...
ð pen and paper
ðĨ digital notes
â writing in full sentences
â bullet points
ðž pictures
ð charts
ðĪŠ funny cartoons?
Invest time finding a learning strategy that works well for you.17:42 PM - 14 Jan 2021
And now for the actual resources I have personally found advantageous in my studies.
Resources
Kaggle micro courses. You can start with Python, Pandas, Machine learning, or any number of others. A great starting point because the mini courses are self contained, but also related to each other. I found these courses extremely helpful for getting practical experience in the skills I've needed to go on. Format: you read a tutorial and then write code in exercises that are checked with tests. Find the courses here.
Deep Learning for Coders with fastai and PyTorch. I'd recommend this book after you have a little bit of exposure to machine learning. Just knowing what it is, and understanding a few basic examples of regression and classification problems before jumping into deep learning was helpful, in my experience. But if you jump right in with this book, I'm sure that could work too. Format has 3 options: physical book, video course, and Jupyter notebooks (the latter 2 are available for free online). There is also an online forum where you can ask questions.
Udacity's Machine Learning course. I have only just started this course. So far I'm enjoying it, and not only for the two instructor's entertaining interactions! This course covers a lot of the theory behind Machine Learning, but not at the expense of remaining very approachable, so it is proving excellent for filling in gaps in my understanding of relevant topics. It's offered by Udacity in partnership with Georgia Tech's online masters degree in computer science program. I may not have the money to pursue that degree, but at least I can start here and earn some of the knowledge that's a part of it for free. Format: video lectures. Here it is!
Bonus: A few other books I have been enjoying:
- Deep Learning with PyTorch: This provides a deeper dive into the PyTorch library along with a good introduction to theory of the various deep learning models being used. Find it here.
- Deep Learning from Scratch: Builds up an understanding of deep learning concepts from the fundamentals, and talks through visual, mathematic, and code examples (in Python) for each. Here it is.
- Data Science from Scratch: If you are more interested in a general overview of data science, before jumping into machine learning, this is a great resource. It takes you through intuitive examples of the practical applications of statistics in understanding data, and also has introduction to machine learning concepts later in the book. Ta-da!
I hope you found this week's machine learning log helpful in some way. What do you think your learning style is? Let's have a conversation in the comments below! :)
Latest comments (4)
Nice! Thanks a lot for sharing. I've done the Kaggle micro courses too (they are really amazing) and was wondering where to go from there.
My learning style is more of a need based learning. I learn concepts as and when the ML model I'm making needs it. But I've hit a plateau now and don't know how to improve the model's accuracy further. So I'm looking for resources.
I can definitely relate to hitting a plateau. I feel like I am starting to emerge from one.
What helped me was beginning to use the skills in a new problem. Looking on kaggle for a dataset to practice on, then doing some exploratory data analysis, building a simple model - generally playing around with the skills you've learned from the micro-courses. Basically do that over and over again, and then look for answers to any questions you have along the way.
I've found it helpful to take notes on the process - it gives you something to refer back to and also documents your progress. :)
Interesting read!
Thank you!