Everybody ‘now’ talks about Neural Networks, Deep learning, and quickly first think that folks do was get some library, put raw data inside e ‘Damn! I am a Data Science.’
Fair enough! However, this is not true, sorry to say that. Know how to use a library is great to small or huge problems but this does not help you when you need to tell your boss when a client will say a goodbye to him in the next few months or predict when one of your millions of costumers will try fraud your system, of course, with a high accuracy or minimum false positives.
Ok, maybe you can ( again, small problems perhaps don’t require a massive neural network to solve it ) and when you get this issue front of you, sorry my friend, you will need to know ‘How this Neural Magic thing works behind the scenes’
Make simple, but not so simple
Many problems can be resolve with small twicks but if you dont know how the basics works, how you can resolve such a think
People get stuck months and not days in calculus, backpropagation, gradients ( ‘partial’ but that is ok, we love generalizing thinks ) just to pass a semester to try to get a Ph.D.
Take shortcuts is not that bad like many people try to show in that way, so let’s do this in the right way ( at least, using common sense )
I will list just three links that will make you more close to a Data Science path and not a ‘Pop/Fake/Hippie Data Science.’
Machine Learning Mastery
Amazing guy, amazing blog! Trust me, many examples, to make you start to understand how a model can use ‘in the right way.’
http://machinelearningmastery.com/
I am Trask
The ‘Real Deal’ to understand how things work from scratch using numpy! Andrew Trask do an excellent job with this book, ( it’s not finished yet, but you can read the first chapters that have a good, understandable content !) and of course, your blog has great posts as well
http://iamtrask.github.io/2016/08/17/grokking-deep-learning/
Luis Serrano
If you like videos to understand thinks, Luis Serrano will make you understand in 40 minutes or less how thinks works just with intuition! Trust me, if you know how to use scikit learn, pytorch, tensor flow BUT don’t have ANY idea was is going one watch this videos!