Learn Deep Learning via Keras examples with absolutely no math
I’m always intrigued when students tell me they want to learn deep learning without doing any math.
I was explaining to someone just yesterday – if you look at <insert famous deep learning book by famous deep learning researcher here> – the entire thing is actually cover to cover equations. Ha!
Anyhow, I wanted to test this hypothesis. How far can one get, if they try to learn deep learning via an API?
So I made this little book. It’s full of Keras examples, starting from a basic feedforward neural network, then adding some modern techniques like dropout and batch norm, then moving to more advanced architectures like CNNs and RNNs.
Of course, if you are a reader of my newsletter, you probably aren’t afraid of math!
But, I thought I’d share this book with you anyway, since it contains some interesting examples that you haven’t seen in my courses before.
– CIFAR dataset
– time series prediction using an RNN
– machine translation using a Bidirectional RNN (not a seq-to-seq model as in my Advanced NLP course)
This would also be a great opportunity to brush up on your Keras skills, which are going to be useful for my next course (hopefully coming out in a few days!)
Finally – I’ve also linked below my related book, “Simple Machine Learning for Programmers” – it is a similar experiment in teaching about machine learning using an API with no math. It’s the same as the machine learning section of my Numpy course but I know some students like to have written versions of things so they can read on the subway / airplane. If so, check it out!