Vector quantization approaches to nonlinear stochastic estimation. Many problems in health, economics, telecommunications and industrial control can be formulated as estimation problems with uncertain data. This project is aimed at developing a novel class of algorithms aimed at high complexity estimation problems. If successful, the project will provide new approaches to these problems.
Joint modelling and recognition of linguistic and paralinguistic speech information. A new modelling framework will be developed exploiting interdependence between linguistic and paralinguistic cues to improve automatic recognition of emotion-related information. Applications in the high-tech industry include automatic routing of angry telephone customers or pre-suicidal crisis centre callers to specialist operators/clinicians.
Automated Diagnosis of Faults in Rotating Machinery using Adaptive Network Based Fuzzy Inference. The long-term integrity of engineering assets depends on the quality of their maintenance which runs into billions of dollars per year in Australia. This project aims to develop a new fundamental automated technique for the detection and diagnosis of machinery faults. The innovation lies in the ability of this technique to not depend on knowledge of fault components in the discrete wavelet packet ....Automated Diagnosis of Faults in Rotating Machinery using Adaptive Network Based Fuzzy Inference. The long-term integrity of engineering assets depends on the quality of their maintenance which runs into billions of dollars per year in Australia. This project aims to develop a new fundamental automated technique for the detection and diagnosis of machinery faults. The innovation lies in the ability of this technique to not depend on knowledge of fault components in the discrete wavelet packet analysis. All other work conducted to date depends on knowledge of these components and their location. The results of this work will vastly improve the costly manually based diagnostics procedures in the maintenance of plant and industrial assets.Read moreRead less
Continuous wave excitation for low power Magnetic Resonance Imaging. This project aims to augment the capabilities of Magnetic Resonance Imaging (MRI) systems, using continuous wave (CW) transmission and signal reception, to image objects using very low excitation power. Any given MRI sequence tries to solve an inverse problem, involving estimation of some subset of hidden states and parameters of the system, given the observed data. Using transient and steady-state CW magnetisation dynamics to ....Continuous wave excitation for low power Magnetic Resonance Imaging. This project aims to augment the capabilities of Magnetic Resonance Imaging (MRI) systems, using continuous wave (CW) transmission and signal reception, to image objects using very low excitation power. Any given MRI sequence tries to solve an inverse problem, involving estimation of some subset of hidden states and parameters of the system, given the observed data. Using transient and steady-state CW magnetisation dynamics to solve inverse problems is expected to advance technology toward lower power, lower cost solutions for MRI scanners in healthcare and industrial applications, including materials science and mineral processing.Read moreRead less
New entropy measures of short term signals for smart wearable devices. This project aims to improve reliability and accuracy of wearable devices by developing a new set of computationally efficient algorithms. Wearable devices can be very effective in remote and continuous monitoring to detect short or bursty anomalous events. Present devices are unable to detect such events effectively due to limited capability in processing short length signal. This project will provide computationally efficie ....New entropy measures of short term signals for smart wearable devices. This project aims to improve reliability and accuracy of wearable devices by developing a new set of computationally efficient algorithms. Wearable devices can be very effective in remote and continuous monitoring to detect short or bursty anomalous events. Present devices are unable to detect such events effectively due to limited capability in processing short length signal. This project will provide computationally efficient algorithms for signal quality analysis and enhanced feature extraction methods in resource constrained wearable devices. This will improve the reliability and performance of wearable devices for adoption in intelligent decision-making systems.Read moreRead less
Modelling of neural plasticity for enhanced performance of brain-machine interfaces. Plasticity of the brain is one of the great scientific challenges of neuroscience. The aim of this project is to model the synaptic changes that occur with reward-modulated spike-timing-dependent plasticity and apply the model to developing plasticity targeted brain-machine interfaces. The significance of this approach is that such plasticity targeted techniques provide the prospect of taking advantage of the un ....Modelling of neural plasticity for enhanced performance of brain-machine interfaces. Plasticity of the brain is one of the great scientific challenges of neuroscience. The aim of this project is to model the synaptic changes that occur with reward-modulated spike-timing-dependent plasticity and apply the model to developing plasticity targeted brain-machine interfaces. The significance of this approach is that such plasticity targeted techniques provide the prospect of taking advantage of the underlying neural plasticity to optimise the form of the neural recording and electrical stimulation. The outcomes will be to greatly improve the performance of brain-machine interface in terms of measures such as the number and sensitivity of channels, as well as robustness and reliability.Read moreRead less
Breathing and snoring sound analysis in sleep apnea. About 800,000 Australians suffer from the disease sleep Apnoea (OSA) which has snoring as its earliest symptom. We develop electronics and snore processing algorithms to classify snorers into OSA-positive and OSA-negative classes, based on advanced technology derived from speech recognition systems.
Biologically Inspired Binaural Coupling for Selective Machine Hearing. This project aims to investigate biologically-inspired binaural coupling models in the context of the deep learning paradigm by formulating desirable higher level auditory structures as neural network sub-systems. This project expects to generate new knowledge for developing the next generation of robust speech processing systems that are capable of mimicking the selecting listening ability of humans when faced with realistic ....Biologically Inspired Binaural Coupling for Selective Machine Hearing. This project aims to investigate biologically-inspired binaural coupling models in the context of the deep learning paradigm by formulating desirable higher level auditory structures as neural network sub-systems. This project expects to generate new knowledge for developing the next generation of robust speech processing systems that are capable of mimicking the selecting listening ability of humans when faced with realistic noisy speech signals and the ‘cocktail party problem’ using innovative binaural feedback systems. This work should provide significant benefits, including improved voice biometrics and selective auditory attention capabilities in machines.Read moreRead less
Integrating biologically-inspired auditory models into deep learning. This project aims to discover how a biologically inspired auditory model can be tightly integrated into a state-of-the-art deep learning speech processing framework, to model, design and verify a deep learning based auditory model. Voice-based technologies, ranging from cochlear implants to smart homes, are growing at a rapid pace and speech interfaces are being integrated with all aspects of our lives. However, there is a gro ....Integrating biologically-inspired auditory models into deep learning. This project aims to discover how a biologically inspired auditory model can be tightly integrated into a state-of-the-art deep learning speech processing framework, to model, design and verify a deep learning based auditory model. Voice-based technologies, ranging from cochlear implants to smart homes, are growing at a rapid pace and speech interfaces are being integrated with all aspects of our lives. However, there is a growing demand to improve these voice-enabled services, making them more secure and less open to cyber-crime attack by unauthorised users. The project is expected to improve techniques for modelling and automatic processing of speech and audio signals, which should provide significant benefits, including improved voice biometrics and cochlear implants.Read moreRead less
Active Sound Control and Noise Cancellation over Space. This project aims to address the critical issues for creating acoustic quiet zones in a noisy environment. It will provide novel signal processing theory for further development of active noise cancellation techniques over spatial regions. New technologies developed from this project are expected to underpin the future development of acoustic signal processing research and will have a broad range of applications such as reduction of noise i ....Active Sound Control and Noise Cancellation over Space. This project aims to address the critical issues for creating acoustic quiet zones in a noisy environment. It will provide novel signal processing theory for further development of active noise cancellation techniques over spatial regions. New technologies developed from this project are expected to underpin the future development of acoustic signal processing research and will have a broad range of applications such as reduction of noise inside cars, creation of individual quiet zones in passenger planes and mitigation of acoustic noise made by industrial plants to neighbouring suburbs. The outcomes from this proposal will also have economic importance as it can reduce the health risk posed to people working or living in noisy environments.Read moreRead less