A Month trying out Machine Learning
I was always interested in Machine Learning but I skipped/procrastinated studying it because of either lack of time or because I was scared of the Maths that this journey might contain. I do like Maths but sometimes if a topic comes up that I haven’t learned yet, I have a difficult time understanding it on my own. But it all changed when I stumbled upon this YouTube channel: Sirajology. His videos were very fast paced and I had hard time understanding the code. I always ended up with the same question, ‘why am I not trying to make Machine Learning programs myself?’
To get started there are very limited resources to get started for a beginner as compared to Web Development, where I had so many `Get Started` courses and tutorials on every topic to choose from.
I began my study with Udacity’s Intro to Machine Learning course, as I already knew college-level calculus, statistics and python. The course has great reviews and many of them claim that it is great course for beginners.
Here is a Review for it:
The learning curve was very smooth and the problems were to be solved at abstract level using sklearn’s functions and classes. It was great experience playing with Enron data-set and finding frauds with just 5–10 lines of code. I also learned a lot of terms like regression, classifiers and feature scaling. Meanwhile I also kept trying Siraj’s tutorials and it was amazing how his code was starting to make a sense and I was able to follow up to code till he used functions that came from Tensorflow, a deep learning library by Google.
Meanwhile I also tried Andrew Ng’s Machine Learning Course on Coursera. Unfortunately it was not using Python, so I didn’t feel comfortable and stopped following it. It contains a lot of information on the inner workings of the algorithms, so I will definitely come back to it in future.
Neural Networks
I was introduced to neural networks from Siraj’s videos but feed-forward and back-propagation were still unknown to me. I wanted to learn about neural networks because “Brain” is a topic that amazes me a lot. I went through many eBooks, articles but all of those were very difficult to understand as they used terms like activation functions and represented the process using mathematical notations which are very hard to read.
Finally I hit this article that is pure gold!
Making a neural network
The article uses words to represent the variables and explains the process step by step. It made me understand Feed Forward and Back Propagation with very little brain-stress. This way I get a basic knowledge of ML and Deep Learning just before the end of last year.
Now looking at 2017, I want to try out Deep Learning course on Udacity and start working on data-sets on Kaggle on my own. I would also like to get familiar with Tensorflow and making deep learning models with it. This would be a great New Year’s resolution. I hope I can make it!
I hope the resources and path I marked in this article can also help others get started that don’t know from where should they begin their journey.
I hope the resources and path I marked in this article can also help others get started that don’t know from where should they begin their journey.
If you liked the article, please recommend, follow and share. Your support means a lot to me. Thanks for reading! Have a nice day!
