Dr Ivan Kiskin
Academic and research departments
Surrey Institute for People-Centred Artificial Intelligence (PAI), Centre for Vision, Speech and Signal Processing (CVSSP), School of Biosciences.About
Biography
Biography
I joined as a lecturer in AI for Multimodal Health Monitoring at the Surrey Institute for People-Centred Artificial Intelligence in January 2022. I obtained my PhD in acoustic machine learning with the Machine Learning Research Group of the University of Oxford, as part of the Autonomous Intelligent Machines and Systems CDT. I obtained a first-class honours MEng degree at the University of Oxford in 2015.
Research highlights
My interests include Bayesian deep learning, acoustical and multimodal signal processing (see Research Interests).
I am currently working on a feasibility study of COVID classification from acoustic data in collaboration with UKHSA and the Alan Turing Institute. A publication highlight includes an overview discussing the limitations of current studies: COVID-19 detection from audio: seven grains of salt [paper]. Further large-scale studies are currently under review.
As part of the HumBug project, I have been leading the development of machine learning acoustic monitoring solutions for mosquito recognition for the purpose of data-driven approaches to malaria prevention. From this collaboration, my most notable achievements are:
- DCASE 2022 Task 5: Few-shot Bioacoustic Event Detection co-organiser [website]
- ACM 2022 Computational Paralinguistics Challenge (ComParE) co-organiser: Mosquito Event Detection [website] [paper] [code]
- Best paper award at NeurIPS 2019 Machine Learning for the Developing World Workshop [paper]
- NeurIPS 2021 dataset and benchmark paper [paper] [reviews]
- Leading a collaboration with the MIDS programme at UC Berkeley, resulting in state-of-the-art performance
PhD opportunities
The Surrey Institute for People-Centred Artificial Intelligence is offering a fully funded PhD in Machine learning for audio with applications for health and wellbeing. Full scholarships to exceptional international students are also available. The position will be hosted jointly in the University of Surrey's Centre for Vision, Speech and Signal Processing (CVSSP) and Faculty of Health and Medical Sciences (FHMS). You will also have an excellent opportunity to partake in the AI 4 Sound project at Surrey.
The PhD will be co-supervised by Dr. Ivan Kiskin, Lecturer in AI for Multimodal Health Monitoring, and Prof. Mark Plumbley, Professor of Signal Processing, EPSRC Fellow in "AI for Sound".
Potential topics are not limited to, but may include:
- Acoustic mosquito monitoring for malaria vector mapping and intervention
- Analysing the role of audio and multimodal data on the effect of sleep; using audio to categorise sleep quality, and thus determine biomarkers for disease onset prediction
- Bayesian deep learning for audio classification and event detection
- The effect of audio compression on ML systems
To learn more about this opportunity, please visit https://www.surrey.ac.uk/fees-and-funding/studentships/machine-learning-audio-applications-health-and-wellbeing
Application Deadline: 22 July 2022, though early applications are encouraged.
Start Dates: July/October 2022
My qualifications
Previous roles
News
In the media
ResearchResearch interests
- Machine learning for audio and signal processing
- Bayesian deep learning
- Machine learning for health
Research collaborations
The HumBug project has developed a novel mosquito survey system to detect and identify different species of mosquitoes by their flight tone using budget smartphones. The project has recently received funding by the Bill and Melinda Gates Foundation allowing us to further advance this innovative system in Sub-Saharan Africa. Ivan Kiskin has led and continues to lead the development of machine learning approaches for the detection and classification of mosquitoes since 2015.
As part of a collaboration between the Turing Institute and Royal Statistical Society, Ivan Kiskin has been working on a feasibility assessment of machine learning approaches for the detection of COVID-19 from acoustic biomarkers. In addition, Ivan has been developing Bayesian Neural Networks to conduct through experiments on data collected through React and Test-and-Trace.
As part of the UKDRI and the Surrey Sleep Research Lab, I am studying the feasibility of using acoustics for sleep quality monitoring, as well as assessing the impact acoustic disturbance has on sleep, and therefore implicitly on long-term wellbeing. This project is undertaken jointly with Prof. Derk-Jan Dijk, Professor of Sleep and Physiology and Director of the Surrey Sleep Research Centre, and Prof. Mark Plumbley, Professor of Signal Processing, EPSRC Fellow in "AI for Sound". Specifically, the scientific objectives are as follows:
- Quantify the feasibility of audio biomarkers for the purpose of sleep quality monitoring
- Incorporate audio sensors with multimodal data for the purpose of measuring the impact of environmental noise disturbance on sleep quality
- Analyse existing acoustic data collected in combination with multiple modalities (EEG, ECG, physical displacement, chest breathing patterns) to categorise distinct phases of sleep
Research interests
- Machine learning for audio and signal processing
- Bayesian deep learning
- Machine learning for health
Research collaborations
The HumBug project has developed a novel mosquito survey system to detect and identify different species of mosquitoes by their flight tone using budget smartphones. The project has recently received funding by the Bill and Melinda Gates Foundation allowing us to further advance this innovative system in Sub-Saharan Africa. Ivan Kiskin has led and continues to lead the development of machine learning approaches for the detection and classification of mosquitoes since 2015.
As part of a collaboration between the Turing Institute and Royal Statistical Society, Ivan Kiskin has been working on a feasibility assessment of machine learning approaches for the detection of COVID-19 from acoustic biomarkers. In addition, Ivan has been developing Bayesian Neural Networks to conduct through experiments on data collected through React and Test-and-Trace.
As part of the UKDRI and the Surrey Sleep Research Lab, I am studying the feasibility of using acoustics for sleep quality monitoring, as well as assessing the impact acoustic disturbance has on sleep, and therefore implicitly on long-term wellbeing. This project is undertaken jointly with Prof. Derk-Jan Dijk, Professor of Sleep and Physiology and Director of the Surrey Sleep Research Centre, and Prof. Mark Plumbley, Professor of Signal Processing, EPSRC Fellow in "AI for Sound". Specifically, the scientific objectives are as follows:
- Quantify the feasibility of audio biomarkers for the purpose of sleep quality monitoring
- Incorporate audio sensors with multimodal data for the purpose of measuring the impact of environmental noise disturbance on sleep quality
- Analyse existing acoustic data collected in combination with multiple modalities (EEG, ECG, physical displacement, chest breathing patterns) to categorise distinct phases of sleep
Supervision
Postgraduate research supervision
The Surrey Institute for People-Centred Artificial Intelligence is offering a fully funded PhD in Machine learning for audio with applications for health and wellbeing. Full scholarships to exceptional international students are also available. The position will be hosted jointly in the University of Surrey's Centre for Vision, Speech and Signal Processing (CVSSP) and Faculty of Health and Medical Sciences (FHMS). You will also have an excellent opportunity to partake in the AI 4 Sound project at Surrey.
The PhD will be co-supervised by Dr. Ivan Kiskin, Lecturer in AI for Multimodal Health Monitoring, and Prof. Mark Plumbley, Professor of Signal Processing, EPSRC Fellow in "AI for Sound".
Potential topics are not limited to, but may include:
- Acoustic mosquito monitoring for malaria vector mapping and intervention
- Analysing the role of audio and multimodal data on the effect of sleep; using audio to categorise sleep quality, and thus determine biomarkers for disease onset prediction
- Bayesian deep learning for audio classification and event detection
- The effect of audio compression on ML systems
To learn more about this opportunity, please visit https://www.surrey.ac.uk/fees-and-funding/studentships/machine-learning-audio-applications-health-and-wellbeing
Application Deadline: 22 July 2022, though early applications are encouraged.
Start Dates: July/October 2022