Deep Learning using Tensorflow and Cloud AI (DLTC)

Holistic view

• Tensorflow Introduction
• Spark vs Tensorflow
• Spark and Tensorflow
• Introduction to Serverless Architecture
• Current Challenges with On-Premise Architectures
• How Google enables higher productivity?
• How Key Google Products fit in Enterprise Architecture?
• How to design modern Data Analytics Pipeline on GCP?
• Hands-on exercise: Getting familiar with Google Cloud Platform

GCP Introduction

• Why Google Cloud Platform (GCP)?
• How Innovations at Google driving Data Engineering and Science globally?
• Key Google Products related to Data and Machine Learning
• The relationship among Data Science and Machine Learning
• Come on same page w.r.t. terms and concepts

Interactive Data exploration using DataLab

• Introduction
• Why Cloud AI
• Google Cloud AI Framework
• Google Cloud AI Layers

Cloud AI

Machine Learning APIs

• Introduction to Machine Learning APIs
• Key ML Use Cases
• Vision API
• Natural Language API
• Translate API
• Speech API

AutoML

• What is AutoML
• Why AutoML
• AutoML using Vision API
• AutoML using Natural Language Processing (NLP)
• AutoML Translation
• Hands-on Exercise(s)

Artificial Neural Network

• Introduction to Neural Networks
• Introduction to Perceptron
• Neural Network Activation Functions
• Basic Neural Nets
• Single Hidden Layer Model
• Single Hidden Layer Explained
• Multiple Hidden Layer Model
• Multiple Hidden Layer Results
• Hands-on Exercise(s)

TensorFlow Introduction

• What is TensorFlow?
• Why Tensorflow?
• Tensorflow vs other Frameworks
• Installing TensorFlow
• History of TensorFlow
• TensorFlow Architecture
• Where can Tensorflow run?
• Introduction to Components of TensorFlow
• Why is TensorFlow popular?
• List of Prominent Algorithms supported by TensorFlow
• Simple TensorFlow Example
• Options to Load Data into TensorFlow
• Create Tensorflow pipeline
• Hands-on Exercise(s)

TensorFlow API

• TensorFlow Graphs
• Variables and Placeholders
• Activation Functions
• Building Models
• Deploying Models on Google Cloud
• Monitoring Model through Tensorboard
• Dropout
• Regularization
• Hands-on Exercise(s)

Keras API

• What is Keras?
• Why Keras?
• Keras Basics
• Working with Keras
• Hands-on Exercise(s)

Transfer learning

• What is transfer learning
• Why transfer learning
• Neural Network Architecture with Transfer Learning
• Hands-on Exercise(s)

Convolutional Neural Networks (CNN)

• CNN History
• Understanding CNNs
• Various Layers like Pooling, Convolution, Relu etc.
• CNN Applications
• Hands-on Exercise(s)

Recurrent Neural Networks

• What are Recurrent Neural Networks?
• Different types of RNNs
• Language model and sequence generation
• Sampling novel sequences
• Vanishing gradients with RNNs
• Gated Recurrent Unit (GRU)
• Long Short Term Memory (LSTM)
• Bidirectional RNN
• Deep RNNs
• Hands-on Exercise(s)

CloudML: Scalable Models on GCP

• Why Cloud ML?
• Running TensorFlow model in Local mode
• Porting TensorFlow models to GCP
• Deploying Models in Production
• Model Predictions
• Hands-on exercise(s)

Conversational AI

• Intro to ChatBots
• Key options available
• Building ChatBot
• Hands-on Exercise(s)

Hands-on Exercise(s):

• GCP
• Machine Learning API
• AutoML
  • Preparing and formatting training data for AutoML Translation
  • Preparing and formatting training data for Natural Language Processing Translation
• Tensorflow
  • MNIST dataset intro
  • ML Introduction
  • Basic Operations
  • Convolutional Network
  • Neural Network Raw
  • Convolutional Network Raw
  • Tensorboard Basic
  • Save Restore model
  • Tensorboard advanced
  • Prevent overfitting with dropout and regularization
• Keras
  • NN
  • Image classification Demo
  • Convolutional net
  • Handwritten Digit Recognition
• ChatBot
  • Amazon Lex Bot
  • DialogFlow Chat Bot
• Deploying Tensorflow Model on CloudML