Introduction to Deep Learning (Black/White version): with complete Python and TensorFlow examples

Description

What is Deep Learning? "How are they called? Neutrons?" Convolutional Neural Networks drive the boom Deep Learning without neurons Neuroscience as a treasure for machine learning About this book Deep Learning: An agile field Exponential growth of interest Acquisition of DL startups Hardware for DL Software for DL The biological role model: The Neuron Your brain - A fascinating computing device Structure of a neuron Signal processing by action potentials Synapses Neuronal plasticity Spike-Timing Dependent Plasticity (STDP) The many faces of a neuron What is the function of a biological neuron? Neurons as spatial feature or evidence detectors Neurons as temporal coincidence detectors Perceptron neuron model Neurons as filters Other neuron models Neural Coding The Perceptron The Perceptron neuro-computer Perceptron learning Perceptron in Python Limitations of the Perceptron Self-Organizing Maps The SOM neural network model A SOM in Python SOM and the Cortex Multi Layer Perceptrons The goal Basic idea is gradient descent Splitting the weight change formula into three parts Computing the first part Computing the second part Computing the third part Backpropagation pseudo code MLP in Python Visualization of decision boundaries The need for non-linear transfer functions TensorFlow Introduction Training a linear model with TensorFlow A MLP with TensorFlow Convolutional Neural Networks Introduction Some history about the CNN model Convolutional and pooling layers in TensorFlow Parameters to be defined for a convolution layer How to compute the dimension of an output tensor Parameters to be defined for a pooling layer A CNN in TensorFlow Deep Learning Tricks Fighting against vanishing gradients Momentum optimization Nesterov Momentum Optimization AdaGrad RMSProp Adam Comparison of optimizers Batch normalization Beyond Deep Learning Principle of attention Principle of lifelong learning Principle of incremental learning Principle of embodiment Principle of prediction Cognitive architectures Exercises Ex. 1 - Preparing to work with Python Ex. 2 - Python syntax Ex. 3 - Understanding convolutions Ex. 4 - NumPy Ex. 5 - Perceptron Ex. 6 - Speech Recognition with a SOM Ex. 7 - MLP with feedfoward step Ex. 8 - Backpropagation Ex. 9 - A MLP with TensorFlow Ex. 10 - CNN Experiments Ex. 11 - CNN for word recognition using Keras Ex. 12 - Vanishing gradients problem Ex. 13 - Batch normalization in TensorFlow