Our future lies in AI/Machine learning, either your products are using them or you don’t exist in Future.
Now rather than giving more theory, lets dive in what you should do to get started with ML:
- Start with Coursera course of Machine Learning by Andwer Ng : https://www.coursera.org/learn/machine-learning/home/welcome
- This will give you idea about machine learning in general
- Basic Machine Learning algorithms
- Maths behind their working, Make sure to code those assignments (will seriously help in learning cutting edge deep networks, if possible do a basic course on linear algebra (not necessary now))
- Theory on parameter setting to train your model
- Also you can try out some sample using scikit learn library (very simple to use interface for getting started, can use https://docs.continuum.io/anaconda/install to install all and get started, you will learn numpy, pandas,pickle to play with data )
- Now to get started with Deep Neural Networks, read this book: http://neuralnetworksanddeeplearning.com/ — very very good book
- By now you cleared the basic concepts and maths behind the scenes, its time to dig deep using latest libraries to get started
- You can start with tensorflow (library by google, can also try other from facebook or microsoft, these libraries enable to write deep neural network and how to run fast on GPU as for big data learning you need GPU)
- To start use this course: https://www.udacity.com/course/deep-learning–ud730
- Go upto lesson 3
- You now know how to use Tensorflow
- Now as there are multiple areas of research like speech, image, nlp and …. I got interested in nlp and here are some pointers for it:
- Deep NLP
- For basics of NLP go through the following stanford course(a little but still helps to understand basics)
- Now for actual deep nlp: this is a must must course : https://www.youtube.com/watch?v=DJHvaGU9SW8&list=PLCJlDcMjVoEdtem5GaohTC1o9HTTFtK7_ , also course material is available here: http://cs224d.stanford.edu/syllabus.html
- This will help you to learn deep nlp
- Now for further research this links compiles other pointers very well : https://github.com/keonkim/awesome-nlp#thought-vector
By now you should be on the cutting edge of the deep NLP, learn more and share to us also.
Concise and to the point article Puneet !
LikeLike