Understanding Neural Networks

Contents:

• Architecture
• Symbols & Naming Conventions
• Flow
• Shapes
• Dense Layer
  • Forward
  • Backward
  • Model
• Activation Functions
  • Sigmoid
  • Stepwise
  • Relu
  • Leaky Relu
  • Softplus
  • Hyperbolic Tangent(Tanh)
  • Softmax
  • Gaussian
• Loss
  • Categorical Cross Entropy
• Optimization Algortihms
  • Gradient Descent
  • Mini Batch Gradient Descent
• What is log
• Regression Intuition
  • Fitting a linear regression model
    • revisiting psuedo inverse
    • revisiting svd and linear systems
    • good old sklearn
  • lets try something with Neural Networks
    • A neural net like perceptron
    • A little bit deep neural net but no activation functions
    • now a neural net with sigmoid applied
    • 2 sigmoids applied in the net
  • lets try with a little bit complex pattern
    • a completely linear model for complex data
    • with a sigmoid introducing non linearity
    • with a relu layer
    • with two relu layers
  • sine wave with a nerual network
    • Trying out with linear model
    • Now with 2 sigmoid layers
    • 6 layers neural network with sigmoid and tanh
• Classification Intuition
  • Basic Classification
    • Logistic Regression with Neural network (sigmoid)
    • 2 sigmoid layers
    • 3 sigmoid layers
    • Something with relu and softmax
  • Easy Spiral Classification
    • 1 sigmoid
    • 2 sigmoids
    • 1 relu layer
    • 2 relu layers
  • Complex Spiral Classification
    • 1 relu layer
    • 2 relu layers
    • A little bit complex model
• Recurrent Neural Network
  • References
  • What is it?
  • Load data
  • How the data looks like
  • Data Preparation
  • Simple RNN
• Basic AutoEncoders
  • Architecture
• Kohonen Self-Organizing Maps
  • References
  • Intro
• https://machinelearningexploration.readthedocs.io
• https://nlpblog.readthedocs.io
• https://recommendation-systems.readthedocs.io

Indices and tables

• Index

• Module Index

• Search Page