For decades, advances in computing have been driven by smaller and faster transistors. As conventional architectures approach their physical and energy limits, the next revolution in computing will require fundamentally new ways of representing, processing, and storing information. Emerging computing paradigms inspired by physics, biology, and mathematics offer exciting opportunities to solve complex problems with dramatically improved energy efficiency.
Can difficult optimization problems be solved by harnessing the natural dynamics of physical systems? Our research explores Ising machines and other physics-inspired computing architectures that map computational problems onto interacting networks of electronic devices. By exploiting collective physical behavior rather than conventional digital logic, these systems provide a promising pathway toward solving combinatorial optimization problems in communication, robotics, logistics, and machine learning.
Can electronic devices emulate the adaptive and parallel information processing of the human brain? Our research explores emerging memory devices and novel materials for neuromorphic hardware that integrates memory and computation, reducing data movement while enabling highly energy-efficient machine learning and intelligent sensing applications.
How can information be represented in high-dimensional spaces to achieve robust, efficient, and explainable intelligence? We investigate hardware accelerators for hyperdimensional computing that leverage emerging electronic devices to perform learning and inference with low computational complexity, high fault tolerance, and exceptional energy efficiency for edge AI applications.
