Mr Shubhr Singh
Email: email@example.comRoom Number: Peter Landin, CS 403
Audio applications of novel mathematical methods in deep learning
Deep learning is now a widespread tool for audio analysis tasks. Most commonly, these use "convolutional" and/or "recurrent" neural networks. But what will be the next generation of deep learning methods? There are many exciting ideas to build on, such as: deep Gaussian processes, capsule networks, gauge equivalent networks, neural ordinary differential equations, normalising flows, and point process networks. This project will explore such possibilities and their usefulness for analysing/generating audio data, and then refine selected methods to create a new generation of audio deep learning that is not only powerful but also insightful. The applications of this can range from automatic music transcription, audio event detection, to understanding the fine details of vocal expression.
C4DM theme affiliation:
Machine Listening, Sound Synthesis
Big Data Processing ()
The module syllabus adopts a hands-on programming stance. In addition it focuses on algorithms and architectures to familiarise students with message-passing systems ((MPI) as adopted by industry. Parallel computing, which implies the simultaneous execution of several processes for solving a single problem, is a mainstream subject with wide ranging implications for computer architecture, algorithms design and programming. The UK has been at the forefront of this technology through its involvement in the development of several innovative architectures. Queen Mary has been involved with Parallel Computing for more than a decade. In this module, students will be introduced to parallel computing and will gain firsthand experience in relevant techniques.