Research

The CSIR Lab seeks to advance AI hardware design for applications in edge intelligence, security, privacy, and human-AI collaboration. Our current research encompasses:

• Neuromorphic Computation: This approach to brain-inspired computation aims to leverage sparse data and foster noise-resilient hardware acceleration. Our research investigates stochastic circuit techniques to reduce both hardware complexity and susceptibility to data breaches.
• Smart Security: Adapting to advanced threats and changing chip/ambient conditions is imperative for security. We focus on developing ML hardware capable of integrating with security systems to support efficient, few-shot incremental learning for enduring security solutions.
• Human-AI Collaboration: In the evolving human-AI ecosystem, it is crucial for AI to transcend its black-box nature, offering decision-making transparency and reliability. Our research delves into probabilistic deep learning and its hardware design implications, enhancing trust and explainability in AI applications.