Understanding Deep Neural Networks Training Course

Part 1 – Deep Learning and DNN Concepts

Introduction to AI, Machine Learning & Deep Learning

• Overview of the history, core concepts, and common applications of artificial intelligence, moving beyond the speculative aspects often associated with the field
• Collective Intelligence: aggregating knowledge shared among numerous virtual agents
• Genetic algorithms: evolving a population of virtual agents through selection processes
• Standard Learning Machines: definition and core principles
• Task types: supervised learning, unsupervised learning, and reinforcement learning
• Action types: classification, regression, clustering, density estimation, and dimensionality reduction
• Examples of Machine Learning algorithms: Linear regression, Naive Bayes, and Random Trees
• Machine Learning vs. Deep Learning: identifying problems where traditional Machine Learning remains the state of the art (e.g., Random Forests & XGBoost)

Basic Concepts of a Neural Network (Application: multi-layer perceptron)

• Review of essential mathematical foundations
• Definition of a neural network: classical architecture, activation mechanisms
• Weighting of prior activations and determining network depth
• Defining network learning: cost functions, back-propagation, Stochastic gradient descent, and maximum likelihood
• Modeling neural networks: structuring input and output data based on the problem type (regression, classification, etc.) and addressing the curse of dimensionality
• Distinguishing between multi-feature data and signals; selecting appropriate cost functions based on data characteristics
• Function approximation via neural networks: theoretical presentation and practical examples
• Distribution approximation via neural networks: theoretical presentation and practical examples
• Data Augmentation: techniques for balancing datasets
• Generalizing results from neural networks
• Initialization and regularization of neural networks: L1 / L2 regularization, Batch Normalization
• Optimization and convergence algorithms

Standard ML / DL Tools

This section provides a comparative overview of key tools, highlighting their advantages, disadvantages, position in the ecosystem, and typical use cases.

• Data management tools: Apache Spark, Apache Hadoop
• Machine Learning libraries: Numpy, Scipy, Scikit-learn
• High-level Deep Learning frameworks: PyTorch, Keras, Lasagne
• Low-level Deep Learning frameworks: Theano, Torch, Caffe, TensorFlow

Convolutional Neural Networks (CNN)

• Overview of CNNs: fundamental principles and key applications
• Core operations of a CNN: convolutional layers, kernel usage
• Padding & stride, feature map generation, and pooling layers. Includes 1D, 2D, and 3D extensions
• Overview of CNN architectures that have set the state of the art in classification
• Image processing models: LeNet, VGG Networks, Network in Network, Inception, and ResNet. Discusses innovations introduced by each architecture and their broader applications (e.g., 1x1 convolutions or residual connections)
• Integration of attention models
• Application to standard classification tasks (text or image)
• CNNs for generation: super-resolution and pixel-to-pixel segmentation. Presentation of
• Primary strategies for enhancing feature maps in image generation

Recurrent Neural Networks (RNN)

• Overview of RNNs: fundamental principles and applications
• Core RNN operations: hidden activations, back propagation through time, and unfolded versions
• Evolution towards Gated Recurrent Units (GRUs) and LSTM (Long Short Term Memory)
• Analysis of various states and the advancements brought by these architectures
• Challenges of convergence and vanishing gradients
• Classic architectures: temporal series prediction, classification, and more
• RNN Encoder-Decoder architectures and the use of attention models
• NLP applications: word/character encoding and translation
• Video applications: predicting the next frame in a video sequence

Generative models: Variational AutoEncoder (VAE) and Generative Adversarial Networks (GAN)

• Overview of generative models and their relationship to CNNs
• Auto-encoders: dimensionality reduction and limited generation capabilities
• Variational Auto-encoders: generative modeling and distribution approximation. Definition and utilization of latent space. The reparameterization trick. Applications and observed limitations
• Generative Adversarial Networks: core fundamentals
• Dual Network Architecture (Generator and Discriminator) with alternating learning strategies and available cost functions
• GAN convergence and associated difficulties
• Improved convergence techniques: Wasserstein GAN, Began, and Earth Moving Distance
• Applications in image/photograph generation, text generation, and super-resolution

Deep Reinforcement Learning

• Overview of reinforcement learning: controlling an agent within a defined environment
• State representation and possible actions
• Utilizing neural networks to approximate state functions
• Deep Q Learning: experience replay and application to video game control
• Policy optimization: On-policy & off-policy methods, Actor-critic architecture, and A3C
• Applications: control of single video games or digital systems

Part 2 – Theano for Deep Learning

Theano Basics

• Introduction
• Installation and Configuration

TheanoFunctions

• inputs, outputs, updates, givens

Training and Optimization of a neural network using Theano

• Neural Network Modeling
• Logistic Regression
• Hidden Layers
• Training a network
• Computing and Classification
• Optimization
• Log Loss

Testing the model

Part 3 – DNN using Tensorflow

TensorFlow Basics

• Creating, initializing, saving, and restoring TensorFlow variables
• Feeding, reading, and preloading TensorFlow data
• Leveraging TensorFlow infrastructure for training models at scale
• Visualizing and evaluating models with TensorBoard

TensorFlow Mechanics

• Data Preparation
• Downloading data
• Inputs and Placeholders
• Building the Graphs
  • Inference
  • Loss
  • Training
• Training the Model
  • The Graph
  • The Session
  • Training Loop
• Evaluating the Model
  • Building the Evaluation Graph
  • Evaluation Output

The Perceptron

• Activation functions
• The perceptron learning algorithm
• Binary classification with the perceptron
• Document classification with the perceptron
• Limitations of the perceptron

From the Perceptron to Support Vector Machines

• Kernels and the kernel trick
• Maximum margin classification and support vectors

Artificial Neural Networks

• Nonlinear decision boundaries
• Feedforward and feedback artificial neural networks
• Multilayer perceptrons
• Minimizing the cost function
• Forward propagation
• Back propagation
• Strategies for improving neural network learning

Convolutional Neural Networks

• Objectives
• Model Architecture
• Core Principles
• Code Organization
• Launching and Training the Model
• Evaluating a Model

Brief introductions to the following modules will be provided, subject to time availability:

Tensorflow - Advanced Usage

• Threading and Queues
• Distributed TensorFlow
• Writing Documentation and Sharing your Model
• Customizing Data Readers
• Manipulating TensorFlow Model Files

TensorFlow Serving

• Introduction
• Basic Serving Tutorial
• Advanced Serving Tutorial
• Serving Inception Model Tutorial