←
Fan Vincent Mo107 posts
Fan Vincent Mo avatar

Fan Vincent Mo

@VincentMo6

Principal Researcher at @Huawei; Alumni at @ImperialCollege @ImperialSysAL; Working in Safety of language models, Privacy-preserving machine learning.

◷ Joined November 2016▧ 8 Photos and videos
Evan Miller avatar
⟳ Fan Vincent Mo retweeted
Evan Miller@EvMill·24 Jul 2023

I hit a bug in the Attention formula that’s been overlooked for 8+ years. All Transformer models (GPT, LLaMA, etc) are affected. Researchers isolated the bug last month – but they missed a simple solution… Why LLM designers should stop using Softmax 👇 evanmiller.org/attention-is-…

Ali Shahin Shamsabadi avatar
⟳ Fan Vincent Mo retweeted
Ali Shahin Shamsabadi@AliShahinShams1·21 Mar 2023

Have you ever wondered how we can salvage federated learning? 27 March at 11:15am GMT, @realhamed will utilize Trusted Execution Environments on clients for local training, and on servers for secure aggregation, to hide model updates from adversaries.

Curious about novel risks that AI introduces for patients and healthcare providers? We're hosting an #AIUK Fringe event to explore Privacy and Fairness in AI for Health – Register now! private-fair-ai.github.io/ Date: 27 March, 10:00-16:30 GMT Location: @turinginst, London 1/6

Akhil Mathur avatar
⟳ Fan Vincent Mo retweeted
Akhil Mathur@akhilmathurs·11 Nov 2022

Great progress towards efficient and adaptive FL on ultra-constrained devices. Work led by @VincentMo6 during his internship at @BellLabs.

Federated Learning for constrained edge devices need to become more efficient and adaptive. In this preprint, we offer Centaur: a framework that adaptively partitions FL training across multiple clients owned by each user. PDF: arxiv.org/abs/2211.04175 1/4

Post attachment
Fan Vincent Mo avatar
Fan Vincent Mo@VincentMo6·1 Aug 2022

Too many attacks breaking the privacy-by-design principle in federated/collaborative learning? All-in-One. Check out our paper analyzing the private information leakage from neural network gradients using information theory. arxiv.org/pdf/2105.13929.pdf

Post attachment
Fadel Adib avatar
⟳ Fan Vincent Mo retweeted
Fadel Adib@fadeladib·29 Nov 2021

Over the past 5 years, I’ve read >1,000 letters of recommendation to grad programs in MIT - mostly EECS, Media Lab, and MechE. Since it's letter-writing season, here is some concrete advice for anyone writing letters based on my experience

Jia-Bin Huang avatar
⟳ Fan Vincent Mo retweeted
Jia-Bin Huang@jbhuang0604·29 Sep 2021

How to cope with paper rejection? Rejection SUCKS! It feels awful that months of hard work did not pay off. 😭 How do we hold a positive outlook when dealing with rejection? A thread of lessons (learned from many rejections) 🧵

Victor Bahl avatar
⟳ Fan Vincent Mo retweeted
Victor Bahl@SuperBahl·28 Jun 2021

@ACMSIGMOBILE MobiSys 2021 has started. Heartiest Congratulations to the author of the two best paper awards this year. Well done! 🍷Now, on to the first keynote by Shree K. Nayar (Columbia) on "Future Cameras: Redefining the Image." Nice start😀

Post attachment
Kleomenis Katevas avatar
⟳ Fan Vincent Mo retweeted
Kleomenis Katevas@minoskt·11 Jun 2021

Check our teaser video of our latest work "PPFL: Privacy-preserving Federated Learning with Trusted Execution" to be presented on June 30th at #MobiSys2021. Preprint (arxiv.org/abs/2104.14380) and Code (github.com/mofanv/PPFL) are both available online.

We create an animation, giving a brief fun introduction of privacy-preserving federated learning + trusted execution environments. Check it out. Based on @ACMMobiSys paper, with @realhamed @minoskt @_EduardMarin_ @Diego_Perino @kourtellis Thanks to @TEFresearch @ImperialSysAL

1:30