Understanding Deep Neural Networks Training Course

Part 1 – Deep Learning and DNN Concepts

Introduction to AI, Machine Learning & Deep Learning

• The history, basic concepts, and typical applications of artificial intelligence, distinguishing reality from the fantasies often associated with the field
• Collective Intelligence: aggregating knowledge shared by numerous virtual agents
• Genetic algorithms: evolving a population of virtual agents through selection processes
• Definition of a standard Learning Machine
• Types of tasks: supervised learning, unsupervised learning, and reinforcement learning
• Types of actions: 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)

• A refresher on essential mathematical foundations
• Definition of a neural network: classical architecture, activation functions
• Weighting of previous activations and network depth
• Defining the learning process: cost functions, back-propagation, Stochastic Gradient Descent, and maximum likelihood
• Modelling a neural network: structuring input and output data according to the problem type (regression, classification, etc.), and addressing the curse of dimensionality
• Distinguishing between multi-feature data and signals, and selecting appropriate cost functions based on the data
• Approximating functions using neural networks: theory and examples
• Approximating distributions using neural networks: theory and examples
• Data Augmentation: strategies for balancing datasets
• Generalising the results of neural network models
• Initialisation and regularisation of neural networks: L1 / L2 regularisation and Batch Normalisation
• Optimisation and convergence algorithms

Standard ML / DL Tools

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

• Data management tools: Apache Spark and Apache Hadoop Tools
• Machine Learning libraries: NumPy, SciPy, and scikit-learn
• High-level Deep Learning frameworks: PyTorch, Keras, and Lasagne
• Low-level Deep Learning frameworks: Theano, Torch, Caffe, and TensorFlow

Convolutional Neural Networks (CNN)

• Introduction to CNNs: fundamental principles and key applications
• Core operations: convolutional layers and the use of kernels
• Padding and stride, feature map generation, pooling layers, and 1D, 2D, and 3D extensions
• Overview of CNN architectures that have defined the state of the art in classification
• Image models: LeNet, VGG Networks, Network in Network, Inception, and ResNet; an analysis of the innovations introduced by each and their broader applications (e.g., 1x1 Convolution and residual connections)
• Implementing attention models
• Application to standard classification tasks (text or image)
• CNNs for generation: super-resolution and pixel-to-pixel segmentation
• Primary strategies for enhancing feature maps for image generation

Recurrent Neural Networks (RNN)

• Introduction to RNNs: fundamental principles and applications
• Core operations: hidden activations, backpropagation through time, and the unfolded version
• Evolution towards Gated Recurrent Units (GRUs) and Long Short-Term Memory (LSTM) networks
• Overview of architectural states and the advancements brought by these models
• Addressing convergence and vanishing gradient issues
• Classical architectures: time series prediction and classification
• RNN Encoder-Decoder architectures and the use of attention models
• NLP applications: word/character encoding and machine translation
• Video applications: predicting the next frame in a video sequence

Generative Models: Variational Autoencoder (VAE) and Generative Adversarial Networks (GAN)

• Introduction to generative models and their relationship with CNNs
• Autoencoders: dimensionality reduction and limited generation capabilities
• Variational Autoencoders: generative modelling, distribution approximation, definition and use of latent space, the reparameterisation trick, and observed applications and limitations
• Generative Adversarial Networks: fundamentals
• Dual network architecture (Generator and Discriminator) with alternate learning strategies and available cost functions
• GAN convergence and common challenges
• Improved convergence techniques: Wasserstein GAN, Began, and Earth Mover's Distance
• Applications in image and photo generation, text generation, and super-resolution

Deep Reinforcement Learning

• Introduction to reinforcement learning: controlling an agent within a defined environment
• Managing states and possible actions
• Using neural networks to approximate state functions
• Deep Q Learning: experience replay and application to video game control
• Policy optimisation: on-policy vs off-policy methods, Actor-Critic architecture, and A3C
• Applications: controlling a single video game or digital system

Part 2 – Theano for Deep Learning

Theano Basics

• Introduction
• Installation and configuration

Theano Functions

• Inputs, outputs, updates, and givens

Training and Optimisation of a Neural Network using Theano

• Neural network modelling
• Logistic Regression
• Hidden Layers
• Network training processes
• Computation and classification
• Optimisation techniques
• Log Loss calculation

Testing the Model

Part 3 – DNN using TensorFlow

TensorFlow Basics

• Creating, initialising, saving, and restoring TensorFlow variables
• Feeding, reading, and preloading data in TensorFlow
• Utilising TensorFlow infrastructure to train models at scale
• Visualising and evaluating models with TensorBoard

TensorFlow Mechanics

• Data preparation
• Data downloading
• Inputs and Placeholders
• Building the Graphs
  • Inference
  • Loss calculation
  • Training operations
• Training the Model
  • Graph structure
  • Session management
  • Training loop implementation
• Evaluating the Model
  • Building the evaluation graph
  • Obtaining evaluation outputs

The Perceptron

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

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
• Minimising the cost function
• Forward propagation
• Backpropagation
• Strategies to improve neural network learning

Convolutional Neural Networks

• Objectives and goals
• Model architecture design
• Core principles
• Code organisation
• Launching and training the model
• Evaluating model performance

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

TensorFlow – Advanced Usage

• Threading and Queues
• Distributed TensorFlow
• Writing documentation and sharing models
• Customising data readers
• Manipulating TensorFlow model files

TensorFlow Serving

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