Deep Learning for Unstructured Data in Keras (DL-UDK)

Week 1

• What is Keras?
• Why use Keras?
• Keras Installation
• Run Keras on GPU
• Configuring Keras Backend
• Google Colab

Week 2

• Introduction to Neural Networks
• Feedforward
• Backpropagation
• Designing a Neural Network from Scratch without any ML Library
• Desiging a basic Neural Network in Keras

Week 3

• Sequential Model
• Functional API
• Define a Problem Statement: Traffic Sign Classification
• Introduction to Layers: Dense, Activation, Dropout, Flatten, Input, Lambda, Output

Week 4

• Introduction to Convolutional Neural Networks
• CNN: Intution, kernels/filters
• Starting with Architectures- LeNet
• Convolutional Layers: Convolution 2D, Cropping 2D

Week 5

• Pooling Layers: MaxPooling2D, AveragePooling2D
• Setting Hyperparameters
• Compile Model
• Fit Model

Week 6

• Image Preprocessing
• Data Augmentation
• Custom Generators
• fit_generator
• Batch Normalisation

Week 7

• Saving Model
• Inferencing on test set
• Callbacks
• Transfer Learning: Introduction

Week 8

• Transfer Learning: Example
• Transfer Learning Models: VGG16, VGG32

Week 9

• Templates for Deep Learning
• Interesting Industry Use Cases
• Projects you can take up

Week 10

• Advanced Deep Learning Architectures