We explore spiking neural networks (SNN) and their promising applicability for low-power computing in applications such as persistent acoustic monitoring.
Explainable artificial intelligence (XAI)
Many AI/ML models including deep neural networks are thought of as black-box models, which makes assessing their performance on test data, and improving the model difficult. We explore approaches of XAI for improving the trust of an end-user, and efficient updating of models in online setting.
Edge-AI
We are implementing neural networks including SNN's, and deep learning algorithms to run on edge devices: field programmable gate array (FPGA) hardware and embedded systems with CPU's and GPU's. We are also developing approaches for AI to run on multiple edge devices and how to effectively communicate the information to remote server(s).
Deep learning applications
For other applications of AI/ML such as for anomaly detection, robotics, IoT and computer vision, visit the pages regarding the research topics and current projects for details.
