Reconfigurable Computing

Mr. do-all architecture.


Reconfigurable computing uses imprecise logic to compute approximately-correct results.

It is more power efficient than precise logics. A naive example is that fixed-point multiplications are approximation of floating-point multiplications.



Approximate computing has been used for image processing, machine learning, deep learning, etc.



My research on approximate computing includes but not is limited to algorithm, circuit, architecture, application.

Part of my work was featured in Qualcomm Innovation Fellowship Finalist 2019.