Artificial intelligence will aid in developing future chips, and future chips will enable better AI, so researchers are working to increase the workloads and lower the costs of machine learning.
Current Research:
- Domain-specific architecture and system design for AI and Machine Learning with emphasis on near-data processing, in-network processing, and co-optimized design using approximate / low-precision computing
- Development of sample-efficient and computationally inexpensive learning methods for deep neural networks with provable generalization guarantees
- Physics-informed stochastic surrogate modeling for high-fidelity simulation; reinforcement-learning-based optimization for manufacturing process optimization; data-driven defective identification and quality improvement
- Theoretical and algorithmic foundations of optimization, machine learning, and statistical signal processing
