Certificate in Deep Learning with TensorFlow

TABLE OF CONTENT

Installation

• Software Installation

Keras Introduction

• Introduction
• Keras Backends
• Design and Compile a Model
• Model Training, Evaluation, and Prediction
• Training with Data Augmentation
• Training with Transfer Learning and Data Augmentation

Scaling Deep Learning Using Keras and Tensorflow

• Introduction to TensorFlow
• Introduction to TensorBoard
• Types of Parallelism in Deep Learning – Synchronous and Asynchronous
• Distributed TensorFlow
• Configuring Keras to use TensorFlow for Distributed Problems

Training, Tuning, and Serving Our Model in the Cloud

• Introduction
• Introduction to Google Cloud Machine Learning Engine
• Datasets, Feature Columns, and Estimators
• Representing Data in TensorFlow
• Quick Dive into TensorFlow Estimators
• Creating Data Input Pipelines
• Setting Up Our Estimator
• Packaging Our Model
• Training with Google Cloud ML Engine
• Hyperparameter Tuning in the Cloud
• Deploying Our Model for Prediction
• Creating Our Prediction API

Setting Up the Deep Learning Playground

• TensorFlow for Building Deep Learning Models
• Basic Syntaxes, Function Optimization, Variables, and Placeholders
• TensorBoard for Visualization

Training Deep Feed-Forward Neural Networks with Tensorflow

• Start by Loading the Imported Dataset
• Building the Layers of the Neural Network in TensorFlow
• Optimizing the Softmax Cross Entropy Function
• Using DNN Predicting Whether Breast Cancer Cells Are Benign or Not

Applying CNN on Two Real Datasets

• Importing the Two Datasets Using TensorFlow and Sklearn API
• Writing the TensorFlow Code to Add Convolutional and Pooling Layers
• Using tf.train.AdamOptimizer API to Optimize CNN
• Implementing CNN to Create a Face Recognition System

Exercise Rnn to Solve Two Time Series Problems

• Understanding the RNN and the Need for LSTM
• Implementing RNN
• Monthly Riverflow Prediction of Turtle River in Ontario
• Implement LSTM Project to Predict Decimal Number of Given Binary Representation

Using Autoencoders to Efficiently Represent Data

• Encoder and Decoder for Efficient Data Representation
• TensorFlow Code Using Linear Autoencoder to Perform PCA on a 4D Dataset
• Using Stacked Autoencoders for Representation on MNIST Dataset
• Build a Deep Autoencoder to Reduce Latent Space of LFW Face Dataset

Generative Adversarial Networks for Creating Synthetic Dataset

• Generative Adversarial Networks for Creating Synthetic Dataset
• Downloading and Setting Up the (Microsoft Research Asia) Geolife Project Dataset
• Coding the Generator and Discriminator Using TensorFlow
• Training GANs to Create Synthetic GPS Based Trajectories