Title
Explainable AI for time series - a signal processing approach
Abstract
State-of-the-art methods for explaining predictions from time series involve learning an instance-wise saliency mask for each time step; however, many types of time series are difficult to interpret in the time domain, due to the inherently complex nature of the data.
I will go over methods where time series explainability is carried out using frequency representations, leaning on established signal processing methodology on signal decomposition. Specifically I will present what a typical learning system for a time-series typically looks like, and then show-case two XAI methods. One that is based on masking-based methods to produce explanations in the frequency and time-frequency domain, and another that uses a bank of bandpass filters to split the time series into frequency bands.
Bio
Tommy Sonne Alstrøm has been an associate professor at the Technical University of Denmark, DTU since 2020. In the last few years, his primary research focus has been on machine learning for signal processing, with applications in spectroscopy and time series analysis for healthcare. These domains share the challenge of modeling sequential data (often sensitive) and adapting to non-stationary conditions, which has driven contributions to federated learning, representation learning, and explainable AI.
Practical
This seminar is organised by the RT 8 Explainability and fairness at Integreat.
The Tuesday seminar series explores a wide range of topics relevant to our research themes. Each week, invited researchers present for 40 minutes, followed by a group discussion.
Seminars are open for attendance for everybody.
For those unable to attend in person: https://uio.zoom.us/j/62535744013
Point of contacts:
- Organiser and moderator: Associate Professor in ML Kristoffer Wickstrøm, University of Tromsø
- Integreat Administrative Leader Maria Dikova, University of Oslo