HardeepArora

[Issam Laradji @ILaradji]

We present a new weakly-supervised method that significantly reduces the cost of labeling of #COVID19 CT-Scans. It trains with a novel consistency loss on point annotations to identify infected regions accurately.
Code: https://t.co/1afMkJMzOb
Paper: https://t.co/yuQigy65TS

[Nicolas Chapados @NicolasChapados]

We are happy to release the code to reproduce the results in our N-BEATS paper (https://t.co/cXAiqrwtXh, presented at the ICLR 2020 conference, available at https://t.co/hK1fGjas0o. N-BEATS delivers SOTA univariate forecasting on many benchmarks. #ElementAI #MILA #forecasting

[Nicolas Chapados @NicolasChapados]

The GPU infrastructure management toolset that Element AI’s researchers have been using internally for 3 years is now a new product. Best way to boost AI research productivity! https://t.co/EF1o1nsRQD

[François Chollet @fchollet]

Rats generalize better than deep RL, and are more sample efficient

[OpenAI @OpenAI]

We've trained an AI system to solve the Rubik's Cube with a human-like robot hand.

This is an unprecedented level of dexterity for a robot, and is hard even for humans to do.

The system trains in an imperfect simulation and quickly adapts to reality: https://t.co/O04izt3KvO

[Mila - Institut québécois d'IA @Mila_Quebec]

Yesterday, Yoshua Bengio gave a lecture on the principles underlying the recent successes of deep learning, its limitations and research directions for advancing AI on a human scale. #AI https://t.co/PjgdnCIQFH

[ServiceNow AI Research @ServiceNowRSRCH]

"Researchers from #ElementAI, @MILAMontreal, and @UMontreal have introduced a powerful transfer language model that can summarize long scientific articles effectively, outperforming traditional #seq2seq approaches." 🤔 👇https://t.co/EkTzUuYSKi

[CFTE @CFTE_Edu]

It was such a pleasure to host @jaydjenkins from @Google and @HardeepArora from @element_ai at Alpha Go screening and panel in #Singapore yesterday!

They discussed with @TramANguyen from #CFTE how AI and big data are shaping industries and people perceptions

#Fall4Fintech

[Google DeepMind @GoogleDeepMind]

Human social cognition begins with our theory of mind: understanding that others have beliefs, desires, and goals. In Machine Theory of Mind, we present a neural network that advances machine social cognition, learning to model what drives other agents. https://t.co/brEEa2hkt5

[Jeff Dean @JeffDean]

Some new work from the part of Google Brain that works on ml for healthcare: we have encouraging early signs that non-invasive retinal images contain subtle indicators of cardiovascular health that ml models can pick up on, that weren't previously even known to human doctors. https://t.co/tWrL7CyuIJ

[HardeepArora @HardeepArora]

@bmurr26 Yup kaggle went down.

[HardeepArora @HardeepArora]

@orcldoug @HemantKChitale @kidrac thanks for arranging another successful SOS. Really useful knowledge sharing.

[HardeepArora @HardeepArora]

@orcldoug @kidrac I agree @JLOracle dedication to community and sharing is commendable.

[HardeepArora @HardeepArora]

Yet another great #SingaporeOracleSessions. Thanks to @orcldoug and @kidrac for pulling this one through.

[HardeepArora @HardeepArora]

Kudos to @JLOracle for taking such fabulous sessions at #SingaporeOracleSessions despite being so ill.

[Doug Burns @orcldoug]

.@JLOracle presenting one of the best things he ever taught me - a long time ago! v$pq_tqstat query See any PX paper I wrote for that

[HardeepArora @HardeepArora]

@karlarao that looks nice, do you think that might be option for us?

[Bjoern Rost @brost]

Does this look like a succesful local user group event or what? Had a great time in #singapore

[HardeepArora @HardeepArora]

@orcldoug @hemantkchitale
Thanks guys for arrange this wonderful session. It was really good and let's have more of this stuff.

[HardeepArora @HardeepArora]

Thanks @TanelPoder for the excellent ExaData troubleshooting master class, I
recommend it anyone wanting to understand ExaData in detail.