Members-Internal

Scaling Deep-Learning Inference with Reconfigurable FPGA-Cluster-based Architecture

Background and Motivation†

↑

Research Goal†

The goal of this thesis is to propose a DNN (CNN) parallelization algorithm, architecture, and design of a Real-time AI-Enabled Real-time Bio-system.

↑

Research Schedule†

DateTask
☑Nov 1 ~ Dec 7, 2020Resize MNIST dataset
Dec 21 ~ TBDUpdate paper "Hand Gesture Processing with Sensor Fusion based on Neuromorphic System"
☑Dec 24 ~ Jan 8, 2021EMG dataset collection using Myo
☑Jan 9 ~ 20, 2021Implementing signal processing part in Matlab
☑Jan 20 ~ Feb 28, 2021Add source code and dataset into github
☑Jan 17 ~ 31, 2021Help to revise ETLTC20210 papers
☑Feb 20 ~ 26, 2021Test 3D printer
☑Mar 2 ~ Mar 5, 2021Study AIRBiS project and write slide
☑Mar 2 ~ Mar 12, 2021Inference of batch test on FPGA and communicate network with UI for 64x64 grayscale
☑Mar 12 ~ Mar 22, 2021Inference of batch test on FPGA and communicate network with UI for 128x128 grayscale
☑Mar 15 ~ Mar 23, 2021Write Hardware Acceleration Part of journal draft
☑Mar 24, 2021Find source code in internet for parallel CNN
☑Mar 26 ~ Apr 2, 2021comparison to other works
☑Apr 3 ~ Apr 14, 2021work on training on FPGA for AIRBiS project
☑Apr 15 ~ Apr 28, 2021Work on speed up for time cost using SDSoC
☑Apr 29 ~ May 16Modify speed up sample code for AIRBiS CNN
☑May 17 ~ June 25Design new CNN mapping algorithm
☑June 21 ~ July 15Prepare and improve the paper for hardware part
☑June 26~ July 7Evaluate CNN mapping algorithm
July 17 ~ August 29Prepare a paper for other publication opportunity
July 17 ~ July 26Optimize convolution acceleration to more speed up
July 27 ~ August 13Design the more secure system on hardware
August 15~ August 31Evaluate the secure system on hardware
Schedule last updated on: 7/19/2021
↑

References†

トップ   新規 一覧 検索 最終更新   ヘルプ   最終更新のRSS