Alvin Wan

Making AI run really fast. Formerly @apple @berkeley_ai

[Alvin Wan @lvinwan]

5/5 - Check out the original paper; if you’re at ICML, find me at Hall 1 #729 today 7/26, 11a-12:30p!
Paper: https://t.co/Y0nwLikFHJ
arXiv: https://t.co/Up4sMxZXWA
Collaboration with: @HanxiangHao, @kaushikpatnaik, Sam Xu, @hadad_omer, @dgueraco, @jrenzhile, Qi Shan

[Alvin Wan @lvinwan]

Optimizing latency & accuracy for neural networks that fit on your phone! Introducing UPSCALE - a new export algorithm for pruned vision models with a 2.1% accuracy boost and up to 2X faster inference speeds! #ICML2023 🧵👇
Website: https://t.co/qv8guBo3hA

[Alvin Wan @lvinwan]

4/5 - For latency, reordering is both beneficial and necessary. Without it, the pruned model is actually 🚨slower🚨 than the *original, unpruned model. Fortunately, reordering reduces latency by 8.6% on average, benefiting even tiny models such as SqueezeNet (-24.9%)📈

[Alvin Wan @lvinwan]

Curious what explainable AI fails to explain? Check it out: https://t.co/PFiPzv3Rhx or learn more about Neural-Backed Decision Trees on our website: https://t.co/eqiqsopumu (just accepted by ICLR 2021)

[Covista Group @covista_ucb]

Do we have to give up privacy to automatically track COVID-19 exposure? Several govt’s around the world have decided “yes,” but there’s an alternative. Just published 👉 https://t.co/XtAhUJaPd6

[Alvin Wan @lvinwan]

@vinayprabhu @TDataScience Thanks! LOL That pic reminds of this unrel dataset: https://t.co/X1yRLTtJrO

[Alvin Wan @lvinwan]

Just published in @TDataScience "What Explainable AI fails to explain (and how we fix that)" https://t.co/eRPmKTfPLb. Feedback welcome!

[Alvin Wan @lvinwan]

@vinayprabhu @TDataScience Oops, thanks for letting me know. It looks like the instance went down because it was overwhelmed. Launching a new instance now with higher capacity.

[Berkeley AI Research @berkeley_ai]

"Explaining what Explainable AI Did Not": @lvinwan describes how to make models as accurate as neural networks but with an interpretable decision process in the new BAIR blog!
https://t.co/nvJois1H8j
#xai

[Alvin Wan @lvinwan]

@19arr87 Thanks! Yep, NBDTs can be used for any classification task, including for NLP. We haven't tried it on NLP ourselves though.

[CSS-Tricks @css]

How to do scroll-linked animations the right way https://t.co/3owxPHaLKx

(As ever, IntersectionObserver is the perfect API for this kind of thing.)

[Smashing Magazine 🇺🇦 🏳️‍🌈 @smashingmag]

In Part 1 of an upcoming article series by @lvinwan, you can learn how to build a simple, interactive VR model for an endless runner game — with webVR and a number of cool A-Frame concepts.

↬ Article: https://t.co/YDGaMSGADG

↬ Demo: https://t.co/Wis3X8f3Vx

[Smashing Magazine 🇺🇦 🏳️‍🌈 @smashingmag]

✏️ An introduction to programming in three dimensions and a guide to deploying a virtual reality model on the web, by @lvinwan:
https://t.co/yBPFH3rMPS

[Alvin Wan @lvinwan]

I'm teaching a new online lesson "Build a Face-Swapping Filter Using Computer Vision" https://t.co/RnpDGZX4tC via @skillshare!

[Smashing Magazine 🇺🇦 🏳️‍🌈 @smashingmag]

✏️ What are the fundamentals of machine learning, and what are the necessary tools to evaluate risk and other concerns in a machine learning application? @lvinwan covers everything you need to get started: https://t.co/4ns72eGJ7o