Deep Learning (Intro)
Machine Learning Vs Deep Learning
Three things that are key: Cloud — GPUs & accessibility to Algorithms
Slideshare (slides)/ 3brown1blue (YouTube link)
How to set up GPU in colab Notebook or Kaggle ?
What are the Datasets to be discussed ?
Churn / Titanic and MNIST
Image of a Neural Network form AstroML (link)
GL (slides)
How does Tensor Board looks (Take example from Deep Learning Book)
What are pre trained NETs
LIVE demo of MNIST (link) with explanation (link)
Kaggle Example on MNIST (link)
Flow with Titanic (link)
Tensor Graph view (link)
What is a Node and What is an Edge, how is it related to a Network Topology ?
Agenda
- Eyes, Nose, Ears and Voice of Robotics
- Discuss Titanic Dataset and introduction to Sweetviz Library
- Discuss about Neural Networks
- Discuss about Tensorflow API (link) (Cheatsheet for Keras (R studio) + Python (link) + CheatSheet explanation (link))
- Introduce MNIST (demo + Notebook)
- 3Blue1Brown Video Demo
- Tensorflow.js LIVE Demos (Any ideas ?)
What are the keywords ?
Number of Neurons
Activation Function
Hidden layers
Loss Function
Optimization Algorithm
No. of epochs
Batch Size
Topics to be covered ?
- Introduction to neural networks
- Activation functions
- Types of activation functions
- Feed forward neural networks
- Basic structure of neural network
- Function of a neural network
- Training a neural network
- Loss function
- Gradient ascent
- Gradient descent
- Back propagation
- Learning rate setting and tuning
- Decay in learning rate
- Momentum
- Introduction to TensorFlow
- Introduction to Keras
- Dropout
