About me

My name is Minh Ho (Full name: Nhut-Minh Ho). I received my PhD in 2020 from School of Computing, National University of Singapore. Currently I'm working as a research fellow at NUS. My thesis topic is generally related to "Optimizing number representation and control errors of approximable programs". My research topics include: approximate computing, energy-efficient systems for deep neural networks and neuromorphic computing, number representations. Recently I shifted my research towards numerical analysis for more general applications including blockchain applications and smart contracts. Besides, I am also investigating the posibility of researching and commercializing AI-assisted code analysis solutions.

Email:
minhhn (at) comp(dot)nus(dot)edu(dot)sg
Profile:
Google Scholar ; Github

Selected publications (full list: Google Scholar)

  1. Nhut-Minh Ho et al. "Bedot: Bit Efficient Dot Product for Deep Generative Models." Submitted for peer review.

  2. Nhut-Minh Ho, and Weng-Fai Wong. "Tensorox: Accelerating GPU applications via neural approximation on unused tensor cores." IEEE Transactions on Parallel and Distributed Systems 2021.

  3. Nhut-Minh Ho, Duy-Thanh Nguyen, Himeshi DeSilva, John L. Gustafson, Weng-Fai Wong, and Ik-Joon Chang. "Posit Arithmetic for the Training and Deployment of Generative Adversarial Networks" 2021 Design, Automation & Test in Europe Conference & Exhibition (DATE). 2021.

  4. Himeshi De Silva, Andrew E. Santosa, Nhut-Minh Ho, and Weng-Fai Wong. "ApproxSymate: path sensitive program approximation using symbolic execution." In Proceedings of the 20th ACM SIGPLAN/SIGBED International Conference on Languages, Compilers, and Tools for Embedded Systems (LCTES) 2019.

  5. Duy-Thanh Nguyen, Nhut-Minh Ho, and Ik-Joon Chang. "St-DRC: Stretchable DRAM Refresh Controller with No Parity-overhead Error Correction Scheme for Energy-efficient DNNs." Proceedings of the 56th Annual Design Automation Conference (DAC) 2019.

  6. Nhut-Minh Ho, Ramesh Vaddi, and Weng-Fai Wong. "Multi-objective Precision Optimization of Deep Neural Networks for Edge Devices." 2019 Design, Automation & Test in Europe Conference & Exhibition (DATE). 2019.

  7. Weng-Fai Wong, Pooja Roy, Rajashi Ray, and Nhut-Minh Ho. "Compilation and Other Software Techniques Enabling Approximate Computing." In Approximate Circuits (2019)

  8. Nhut-Minh Ho, and Weng-Fai Wong. "Exploiting half precision arithmetic in Nvidia GPUs." In 2017 IEEE High Performance Extreme Computing Conference (HPEC) 2017. Best paper finalist

  9. Nhut-Minh Ho, Elavarasi Manogaran, Weng-Fai Wong, and Asha Anoosheh. "Efficient floating point precision tuning for approximate computing." In 2017 22nd Asia and South Pacific Design Automation Conference (ASP-DAC), 2017.

Selected Open-source Tools

  1. Trustless NFT Migration, 2021 : A full-stack solution and smart contracts for trustless NFT migration across multiple blockchains. It received the second prize at Singapore Blockchain Innovation Challenge 2021
  2. QPytorch+, 2020 - : Enabling low bitwidth Posit arithmetics and novel number formats for the training and inference of neural networks on Pytorch. Published Python package received >2000 installations: PIP download statistics
  3. Tensorox, 2020 : Using low-level CUDA tensor API to implement small neural networks to approximate and accelerate compute intensive CUDA applications
  4. FpTuning, 2016 : A distributed search algorithm utilizing MPI to determine how many bits are required for the fraction of each floating point variable in a program. It was used by Open Transprecision Computing project (http://oprecomp.eu/)

Misc

I was a freelance smart contract developer and blockchain engineer for a short period in 2017. I am also a full stack nodejs web developer for some of my personal projects that are not released yet. Some of the below are simple experimental scripts and website

Covid Tracker for busy people ?: Link

Experimental scripts on website (petals falling from static picture, does not support smartphone's browsers): Link