Neural Networks
This module introduces the theory and practice of neural computation.
Neural Networks are widely used techniques for modelling and classifying data. They are used in industry for data analysis applications such as image classification, speech analysis and regression tasks.This module offers the principles of neuro-computing with neural networks widely used for addressing real world problems such as regression, pattern recognition and time-series prediction.
Topics covered
- Introduction to Connectionist Learning
- Implementing Learning Algorithms for Single-Layer Perceptrons
- Multilayer Perceptrons: On-line and Batch Backpropagation Algorithms
- Radial-Basis Function Networks
- Neural Network Tuning: The Bias/Variance Dilemma
- Overfitting Avoidance
- Unsupervised Learning with Self-Organizing Kohonen Networks
- Deep Learning
- Hopfield Type Recurrent Neural Networks
- Applications of Neural Networks
Credits
15 (150 hours)
Assessment
- Coursework item 1 (50%)
- Coursework item 2 (50%)