Neural networks by example
summary
Explains how simple networks and convolutional nets actually get built and trained on MNIST, using a perceptron and a CNN with convolution, pooling, ReLU and fully connected layers and trained by backprop to tune weights, showing ~92% accuracy in Keras/TensorFlow and ~90% in CNTK, runs on a Titan GPU, and deploys a saved model in a C# web app that classifies uploaded images MNIST 28x28 grayscale TensorFlow with Keras Microsoft CNTK C# web demo
AI-written summary · about
chapters (12)
- 0:00Introduction
- 7:34Neural networks history
- 15:00Neurons
- 17:10Neural Networks
- 21:03Simple Neural Network
- 23:25Hidden Neural Network
- 25:34Neural Networks Types
- 29:29Convolutional Networks
- 31:05Feature Map
- 36:49Fully Connected
- 40:22Convolutional
- 42:48NIST Case Study
related talks
01h 04mWriting a Neural Net from Scratch
01h 50mNeural Networks: Walkthrough
58mNeural Networks: A Primer
37mCode First, Math Later: Learning Neural Nets Through Implementation and Examples
01h 02mNeural Networks: A Primer
26mMachine Learning in the Browser with Deep Neural Networks
27mDemystifying Deep Neural Networks
30mMachine Learning & Neural Networks Demystified