The aims of this module are: To instill an understanding of the foundations, the applied design, and architectures of artificial neural networks. To be able to apply these techniques to deep learning networks. To be able to apply and to defend the design choices of the model, based on best practices. To be able to evaluate (using an ethical lens) and deploy neural networks for production.
Artificial Neural Networks
Developing artificial neural networks, understanding of the perceptron, weights and activation functions. Training an artificial neural network using applied techniques such as backpropagation. Evaluation of network topologies (capacity and depth), learning rates, optimisers, epochs and batch sizes.
Deep Learning
Evaluate and understand multiple deep learning models. Applying models to classification and regression contexts. Serialization of a deep learning model. Apply convolutional neural networks (CNN), for image recognition and Large Language Models (LLMs) for natural language processing (NLP) contexts. Apply recurrent neural networks (RNN) for time series prediction.
Model Performance
Evaluate and improve model performance by hyper-parameter turning with methods such as grid searching, checkpointing of models and plotting model history. Examine generalisation of models (under and overfitting) and apply regularisation techniques. Develop ethical and robust artificial neural network models. Deploy neural networks for inference.
| Module Content & Assessment | |
|---|---|
| Assessment Breakdown | % |
| Formal Examination | 50 |
| Other Assessment(s) | 50 |