Bio-inspired pre-processing to improve object and/or speech classification

Project description

Modern image and speech classification performs well, but still encounter difficulties, particularly in cluttered and poorly/unevenly lit environments and amongst noise or people with uncommon accents. Most research focuses on developing better classification algorithms, with less effort devoted to improving the quality of the incoming signal. Application of bio-inspired pre-processing of visual images has shown improvement in the ability to detect items of interest, but no work to date has looked at the quality of classification of these objects. Furthermore, these techniques have not yet been applied to acoustic signals for speech recognition purposes. This project will see if neural networks do a better job of object and/or speech classification if the incoming data have first been filtered by an existing model of the biological vision and/or acoustic system.

Supervisors research focus

My goal is to bring robotics out of the lab and into the real-world. To stop having to adapt our environments to suit artificial systems and build artificial systems with the ability to adapt to different environments. I will continue to contribute to the technological and scientific progress of Australia both directly and by helping to guide future generations of entrepreneurs and engineering professionals. My research focus is biologically inspired sensors and signal processing. Data is everywhere; information is the lifeblood of the modern age, but the refinement of raw data into usable information is a complex task. Traditional approaches have tended to be linear and time-invariant, or at least deviate minimally from this paradigm. Adaptation is the hallmark of biological systems, which, by definition, is non-linear and time varying. By studying how biology works, and putting that processing into action, we can open up a new world of possibilities in many areas of technology.

Note: You need to register interest in projects from different supervisors (not a number of projects with the one supervisor).
You must also contact each supervisor directly to discuss both the project details and your suitability to undertake the project.