Deboleena Roy

ML Research Scientist @Meta | ml, comics and sci-fi

[mickey friedman @mickeyxfriedman]

patient 0 of LLM psychosis

[BuccoCapital Bloke @buccocapital]

Why stop here? Hear me out. A single BNPL contract for a burrito is incredibly risky. Anyone willing to pay for a burrito in installments can't be trusted to pay their debt. But what if...we pooled the payments together? The risk would go away.

[Tanvi @tanvishares]

Excited to talk about my work on “What, When, Where to Compute-in-Memory” today at DAC. Please stop by Level 2 Lobby between 7:00pm and 9:00pm if you are also attending! The work explores key questions when integrating compute within on-chip memory for ML
@61stDAC #DACYoungFellow

[AI at Meta @AIatMeta]

Introducing the next generation of the Meta Training and Inference Accelerator (MTIA), the next in our family of custom-made silicon, designed for Meta’s AI workloads.

Full details ➡️ https://t.co/bF9tn4TfeJ

[Latest in space @latestinspace]

BREAKING 🚨: India's #Chandrayaan3 space probe has successfully landed on the Moon

The first country ever to visit the lunar south pole

[Hank Green @hankgreen]

Massively under-reported science story because there's so much going on right now but...it turns out that we might have figured out what's causing this very scary spike.

Quick thread, on how WE'VE BEEN ACCIDENTALLY GEOENGINEERING FOR DECADES...but then we stopped:

[WeRateDogs @dog_rates]

In honor of #NationalPuppyDay, please reply with a pic of your puppy (every dog is a puppy) ❤️

[PBF Comics @PBFcomics]

The Myth of Sisyphus

[Yann LeCun @ylecun]

FAIR played a key role in making the AI R&D scene open.
Others followed. At least for a while.

Now, OpenAI, DeepMind, & perhaps even Google are clearly publishing & open-sourcing considerably less.

What will be the consequences on the progress of AI science and technology?

[Deboleena Roy @deboleenaroy]

Congratulations @gobindasaha07

[Gobinda Saha @gobindasaha07]

Our paper "Saliency Guided Experience Packing for Replay in Continual Learning" was selected as an Award Finalist paper at WACV 2023 :):). Paper link: https://t.co/DXuW82vtl4
@LifeAtPurdue #wacv2023 @wacv_official

[The Paperclip @Paperclip_In]

Over 700 years ago, a nomadic tribe from Rajasthan realized that there was no fixed temple they could visit with their constant roving. They, thus, created temples that could travel with them which subsequently introduced one of the richest artforms of the country. A thread. 1/10

[Joe Weisenthal @TheStalwart]

Wow, check out the timing of the text that California sent out to the public, and then the moment that electricity consumption in the state started to roll over.

https://t.co/Kgl9TwZGom

[jeffrey hui @jeffistyping]

too academic for industry, too industrial for academia, just in time to work in machine learning engineering

[C-BRIC @CBRIC_Center]

C-BRIC researchers Adarsh Kosta and Chankyu Lee from @PurdueECE presented their work at the ICRA conference. More at https://t.co/vZp4KaDt9R

[John Collison @collision]

As you become an adult, you realize that things around you weren't just always there; people made them happen. But only recently have I started to internalize how much tenacity *everything* requires. That hotel, that park, that railway. The world is a museum of passion projects.

[The Sarcastic Jerk @The_Sarcastic_J]

Next Batman movie will be a podcast.

[Rajan Patel @rajanpatel]

Fun fact. 15% of the searches we see every day have never been Googled before.

[Yann LeCun @ylecun]

Any function (trained or not) that maps multiple different inputs to identical or similar outputs is subject to adversarial examples.
It's particularly true of functions with high-dim inputs and discrete outputs, like image classifiers.
Nothing to do with DL or CNN specifically.

[Sam Bowman @sleepinyourhat]

PhD admissions season is ramping up, so I feel obliged to join the chorus of voices reminding everyone that doing a PhD is, in most cases, a terrible idea.