Understanding how the brain encodes, stores, and retrieves information would not only help detect and cure brain-related diseases, promote recovery from injury, but also provide a basis to preserve cognitive function under challenging circumstances. In the meantime, cognitive computing system may provide solutions to the explosive increase in big data computing needs, and the growing gap in what conventional Von Neumann computer systems can deliver.
The integrated Neuromorphic Electronics and MicroSystems (iNEMS) lab is a collaborative effort that brings together researchers in nanodevices, computer architecture, neural science and machine learning. In one research theme, we aim to build a new neural interface system that can record and decipher how information is encoded and processed in the brain during learning, store this information in electronics for future use, and possibly send this information back to brain circuits to in a closed loop. Potential applications include neuroscience studies and function augmentation. In another theme, we aim to build brain-inspired, efficient cognitive computing hardware systems. Our envisioned computing system is drastically different from the conventional Von Neumann paradigm and will exploit sparse neuron spikes to achieve significant advances in both performance and energy efficiency, much like how the brain works. We combine our expertise in emerging devices and efficient circuit and architecture design in building the brain-inspired computing hardware. Potential applications include hardware acceleration of neuroscience simulations and other big-data type problems including video, audio processing and sensor fusion.