Jiahe Shi

I am a master student advised by Prof. Jun Tao at Fudan University, where I received my B.S. degree. I am very lucky to work with Prof. Zhiru Zhang on ML for EDA at Cornell University, Prof. Yiyu Shi on Federated Contrastive Learning at the University of Notre Dame, and Prof. Peng Li on statsitical model for EDA at University of California, Santa Barbara. (chronically)

My research interest is at the intersection of machine learning and system/hardware design.

Email  |  LinkedIn  |  Google Scholar

The first work that enables learning visual representations from unlabeled streaming data on resource-restricted distributed devices.

Developed an algorithm to automatically optimize the capacitor array in a wideband VCO.

A novel Bayesian Inference method based on Bernoulli distribution with conjugate prior to efficiently estimate parametric yields over multiple PVT corners.

An efficient Bayesian optimization method based on Bayesian neural network for yield optimization in multiple environmental corners.

A new neuromorphic learning and computing approach for electromyogram (EMG) based hand gesture recognition tasks.

• The player can take control of an aircraft to defeat the enemy aircraft and save the Earth.
• Single-player and co-op modes are developed and can be selected by the player.
• The game consists of five stages with increasing difficulty, various projectiles, background music, and action audio effects.
• Various functions are implemented, including pause, exit, customized setting, and help.

• An ARM processor reads pre-saved images from DDR, then processes them through neural networks implemented in FPGA. The recoginition results are displayed on a screen via a VGA port.
• We used Vivado HLS to design the IP cores of resize, convolution, and pooling.
• We constructed a display data path using Vivado IPs.
• We assigned some DNN computing to an ARM processor on the Zybo Z7 board due to limited computation resource.