You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
This is an implementation of different optimization algorithms such as: - Gradient Descent (stochastic - mini-batch - batch) - Momentum - NAG - Adagrad - RMS-prop - BFGS - Adam Also, most of them are implemented in vectorized form for multi-variate problems
A repo holding the implementation as well as some theoretical explanation of the important relevant concepts. It is going to be in development for a long long time. I'll keep adding things everytime I have something to add to it, and I have the time for it. One can use it to learn the basics of Machine Learning from kind of scratch.
Tensorflow implementation of asyncronous 1-step Q learning in "Asynchronous Methods for Deep Reinforcement Learning" with improvement on weight update process (use minibatch) to speed up training.