Applied Deep Learning for Engineers (ELEN90099)

This subject covers a modern deep learning approach to engineering using a project-centric pedagogy.  Building upon system optimisation and machine learning fundamentals presented in ELEN90088, the subject will present advanced deep learning architectures to address long-standing engineering challenges such as system complexity, curse of dimensionality, and modelling gap. Subject will specifically focus on engineering problems from multiple application areas including Internet of Things (IoT), smart grid and power systems, robotics, cyber-security, and communication networks. The concepts taught in this subject will lead to a better understanding of how advanced deep learning frameworks can be applied to modern engineering and cyber-physical systems.

INDICATIVE CONTENT
Topics covered may include:

• Latent spaces, auto encoder architectures.
• Advanced deep learning architectures, auto-differentiation, physics-inspired neural networks.
• Sequential data analysis and predictive models such as transformers.
• Generative models such as GANs and GPT variants.
• Other advanced topics such as meta parameter optimisation, Markov Chain Monte Carlo sampling.
• Distributed machine learning, federated learning, graph neural networks.
• Cyber-physical security of modern engineering systems, including data-based anomaly and threat detection and prediction.

Subject projects will focus on engineering applications in areas such as Internet of Things (IoT), smart grid and power systems, robotics, cyber-security, and communication networks.