Latest Module Specifications
Current Academic Year 2025 - 2026
No Banner module data is available
| |||||||||||||||||||||||||||||||||||||||||||||||||
|
Description This module will provide students with both a theoretical and practical grounding in modern machine learning. The module includes classical approaches for the purposes of historical context, conceptual development, interpretability and specific solution applicability. However, the primary emphasis is deep learning and contemporary approaches such as reinforcement learning. Both supervised and unsupervised contexts will be explored. The educational structure is based on a project-oriented approach with a mix of both group and individual activities. Students will gain hands-on experience using research-grade datasets, international challenges and practical tutorial sessions. | |||||||||||||||||||||||||||||||||||||||||||||||||
|
Learning Outcomes 1. Demonstrate a systematic understanding of representation concepts in machine learning contexts 2. Design, critique and analyse a machine learning system suitable for a given problem with due consideration given to performance metrics, class/data skew and explainability requirements. 3. Evaluate complex solutions from open science solution repositories to address a sequential data challenge 4. Design, train and evaluate a machine learning solution for an open and relatively ill-defined custom research challenge introduced by instructor based on research being carried out at DCU at PhD level and beyond and reported in the form of a scientific short paper. 5. Demonstrate a systematic understanding of modern machine learning techniques such as deep neural networks, graph- neural networks and reinforcement learning approaches. 6. Select, evaluate, adapt and deploy complex solutions to address a multimodal machine learning challenge | |||||||||||||||||||||||||||||||||||||||||||||||||
| |||||||||||||||||||||||||||||||||||||||||||||||||
| |||||||||||||||||||||||||||||||||||||||||||||||||
|
All module information is indicative and subject to change. For further information,students are advised to refer to the University's Marks and Standards and Programme Specific Regulations at: http://www.dcu.ie/registry/examinations/index.shtml |
|||||||||||||||||||||||||||||||||||||||||||||||||
|
Indicative Content and Learning Activities
Feature Engineering The importance of features in machine learning including selection, synthesis, extraction. To include feature free approaches. Automatic feature development will be introduced here as well as representation learning. Specific approaches for signal processing (time series), images and video and documents will be considered. Classical Machine Learning Logistic Regression, Support Vector Machines, Perceptron, Shallow Neural Networks, Deep Neural Networks Backpropagation, convolutional neural networks, autoencoders, recurrent neural networks, deep neural network architectures, graph neural networks. Cross Cutting Content Automatic Machine Learning, Ensembling, Ethics, Interpretability, Explainability, Large Scale Machine Learning, Practical Implementation considerations, performance metrics, class and data skew, evaluation, model selection Ensemble Methods Bagging, boosting, stacking Reinforcement Learning Concepts, policy-based, value-based, and actor-critic methods | |||||||||||||||||||||||||||||||||||||||||||||||||
|
Indicative Reading List Books: None Articles: None | |||||||||||||||||||||||||||||||||||||||||||||||||
Other Resources
| |||||||||||||||||||||||||||||||||||||||||||||||||
| New code needed for MCM structures for level 8 module CA4015. No change from CA4015 | |||||||||||||||||||||||||||||||||||||||||||||||||