Computer Science and Engineering
Master's Degree
| Status | Complete |
|---|---|
| Part Time | No |
| Years | 2007 - 2009 |
Dr Salem Ameen's Qualifications (2)
Machine learning
Doctor of Philosophy
Status Complete Part Time No Years 2013 - 2017 Project Title Optimizing deep learning networks using multi-armed bandits Project Description This project investigates advanced pruning methods in deep learning to optimize neural network models for reduced size and computational requirements without sacrificing accuracy. The focus is on developing new algorithms using multi-armed bandit methods, including Epsilon-Greedy, Upper Confidence Bounds (UCB), Thompson Sampling, and EXP3. These algorithms were rigorously tested against traditional neural network models and various learning methods. Results indicate significant improvements in model efficiency and accuracy, demonstrating the potential of these pruning techniques in enhancing the practical application of deep learning models. Awarding Institution The University of Salford Director of Studies Sunil Vadera Thesis Optimizing deep learning networks using multi-armed bandits