Junbo Ma

Citation

Thesis Title
Machine Learning and Audio Processing

Read article at Massey Research Online:

The recent progress in artificial intelligence has brought many new opportunities to change the way humans live. Such technological development is supported by innovation and fundamental research in machine learning. However, existing machine learning approaches are still faced with many challenges, from model interpretability to effectiveness in handling data complexity. Motivated by the challenges, Mr Ma developed three highly innovative approaches. The first and second address two crucial theoretical issues in deep learning and multi-view clustering, which are the two fundamental subfields in machine learning. The third approach was developed for polyphonic sound event detection, which is one of the most important applications in audio processing. Both theoretical derivations and experimental results demonstrated their effectiveness and efficiency. These developed approaches address the frontier in machine learning theories and practical problems in audio processing and will generate high impacts in both academia and industry.

Supervisors
Professor Ruili Wang
Dr Andrew Gilman
Dr Seyed Reza Shahamiri
Professor Xun Wang
Dr Helen Zhou