Amongst the AI subfields, machine learning is proliferating in diverse application domains such as cybersecurity, manufacturing, bioinformatics, and healthcare. Fundamental advances in machine learning alongside computer vision and knowledge reasoning are germinating a record number of new algorithms. Such rapid growth in this field has created a significant knowledge gap for non-native AI users. MATRIX researchers will mitigate these gaps by investigating methods for:
- Model identification and validation
- Data pre-processing
- Training neural networks for imbalanced and heterogeneous data
- Computationally light designs for deployment on the edge (eg: IoT, sensors)
- Scalable algorithms that meet performance and application demands in real-time
- Collaborative and federated learning
This group will also design consultation models for industry, community datasets for targeted domains, and collaborative frameworks for community engagement. Targeted domains of immediate interest include healthcare, defense, and cities.