Introduction to Deep Learning

Overview

Learn how to mathematically design, interpret and evaluate deep learning models that are designed to overcome challenges in traditional machine learning models by learning successive layers of representations in terms of other simpler representations, modelled as neural networks.

Course Details

This course provides an in-depth theoretical introduction to Deep Learning - a subfield of machine learning that is underpinning modern advances in artificial intelligence with applications including self-driving cars, object detection, machine vision, speech recognition, fraud detection, drug discovery and bioinformatics. This course explores in detail the major topics in Deep Learning including gradient-based optimisation, deep feedforward networks, regularisation and optimisation, recurrent neural networks and long/short term memory networks, convolutional neural networks, deep belief networks, generative adversarial networks and autoencoders. This course is a fundamental pre-requisite in order to design, build, interpret and evaluate applied deep learning models using modern deep learning frameworks including TensorFlow and Keras, as it enables principal and lead data scientists to genuinely understand how deep learning works under-the-hood beyond simple deployment of existing cloud-based deep learning services.

Course Modules

• 1. Introduction to Deep Learning
• 2. Neural Network Fundamentals
• 3. Gradient-Based Optimisation
• 4. The Perceptron
• 5. Deep Feedforward Networks
• 6. Regularization and Optimisation
• 7. Recurrent Neural Networks Part 1
• 8. Recurrent Neural Networks Part 2
• 9. Convolutional Neural Networks Pt1
• 10. Convolutional Neural Networks Pt2
• 11. Deep Belief Networks Part 1
• 12. Deep Belief Networks Part 2
• 13. Generative Adversarial Networks 1
• 14. Generative Adversarial Networks 2
• 15. Autoencoders Part 1
• 16. Autoencoders Part 2

Requirements

• Linear Algebra or equivalent.
• Statistical Learning or equivalent.
• A-Level Mathematics or equivalent as a minimum, ideally 1st Year Undergraduate Mathematics (modules in Probability, Statistics, Linear Algebra and Analysis) or equivalent.

Outcomes

• Knowledge of the major mathematical topics in Deep Learning.
• Knowledge of modern deep neural architectures and their typical applications.
• The ability to select and train neural models based on the specific context of the business problem.
• The ability to mathematically design, build, interpret and evaluate deep learning models.
• Foundational mathematical knowledge required to design and build applied deep learning models using relevant software frameworks (e.g. TensorFlow and Keras) for modern artificial intelligence use-cases.