Post has attachment

Post has attachment

Post has attachment

Post has attachment

The 2nd edition of my book 'Practical Machine Learning in R and Python:Second edition" is now available in Amazon in paperback ($10.99) and in kindle versions($7.99). The 2nd edition has more content, detailed comments and much better formatting for easier readability. Pick up your copy today!! #machinelearning #r #python #amazon

Post has attachment

Post has attachment

Post has attachment

Post has attachment

Take a look at my latest post 'Deep Learning from first principles in Python, R and Octave-Part 4', in which I implement a simple 2 layer Neural Network with the Softmax as the output unit. In this post I also derive the "Jacobian' of the Softmax, the cross entropy and gradient derivations required for back propagation. This is then implemented in vectorized Python, R and Octave on the Spiral data set (CS231n). The neural network traces the non-linear spiral boundaries between the 3 classes. Check it out! #deeplearning #jacobian #softmax #kroneckerdelta #crossentropy

Post has attachment

Check out the 3rd and latest part in my series on Deep Learning 'Deep Learning from first principles - Part 3'. In this part I implement a multi-layer Deep Learning network with arbitrary depth and height, and with relu, tanh and sigmoid activation functions, in vectorized Python, R and Octave.. The Deep Learning networks generate the necessary non-linear boundaries between 2 datasets a) 9 random clusters and b) Bernoulli's lemniscate. I also perform a simple binary classification using the MNIST dataset of handwritten digits, which according to Prof Geoffrey Hinton is the "Drosophila of Deep Learning" . Take a look! #DeepLearning #R #Octave #Python #lemniscate #MNIST #multilayer

Post has attachment

Wait while more posts are being loaded