Signal Processing Systems (SPS)
Main research interests
Investigating fundamental aspects of data science from the perspectives of Probabilistic Modeling and Information Theory. Applications to streaming data processes such as medical and multimedia signal processing systems, and communications systems. Specific research interests include:
- Universal data compression technology for data modeling. Models can be widely applied to solve problems such as anomaly detection, classification and forecasting. Our research builds on the Context-Tree Weighting algorithm, which is a best-in-class modeling algorithm that was invented in SPS.
- Bayesian machine learning technology to support data-driven design and personalization of wearable computing systems (http://biaslab.org).
- Advanced Communication Techniques, e.g. for wireless transmission in automotive settings. Intelligent Lighting systems, that have, besides illumination, other functionalities (e.g. communication, sensing, positioning). Authentication and Identification Techniques based on biometric modalities but also on hardware intrinsic properties (PUFs). Sensing, Mapping and Localization for highly automated driving. Medical Information Processing (patient data, signals, images, video).
The group is a key player in the broad Eindhoven innovation eco-system, with strong links to key companies and hospitals. As such the group is also a major prosumer of massive data sets and streams in the above application areas.
Our Context-Tree Weighting (CTW) algorithm is a universal data compression algorithm for the class of tree sources that combines an excellent performance with a straightforward analysis. In an information-theoretical sense it is optimal. It received the IEEEE Information Theory Society Best Paper Award.
- The SPS staff includes 5 IEEE Fellows and 10 part-time full professors with joint appointments in industry or clinical institutes. Six staff members are advisors in industry.
- The SPS group has spun out 9 startup companies: Medecs, CED, ViNotion, Iphion, Livassured, Medsim, Nemo, Hipermotion, 3Sense Innovations.
- HearScan A data-driven approach to hearing aid design. STW, GN Hearing (https://goo.gl/XTPtHM)
- ENIAC, Energy Efficient and Intelligent Lighting Systems, e.g. Supervised Learning Through Feature-Based Models, PhD thesis by Amir Jalalirad.
- PATRIOT Eurostars project. PUFs: Anchors of Trust in Resource Constrained Environments.
- i-CAVE, Integrated Cooperative Automated Vehicles. STW Perspectief Project.