Bhav Ashok

Understanding robotics data @RoboticsData

[AK @_akhaliq]

Block-NeRF: Scalable Large Scene Neural View Synthesis
abs: https://t.co/xDa1bmJaCJ
project page: https://t.co/IQTnknxTuK

[Bhav Ashok @bhavashok]

Anyone who has a sufficiently well developed imagination should be able to appreciate that we no longer live in a world where humans have a monopoly on creativity. If you thought otherwise before, this should change your mind.

[Ryan Moulton @moultano]

Come, walk with me for a while through latent space.
https://t.co/MbtvMyBB3t

[Google DeepMind @GoogleDeepMind]

Introducing NFNets, a family of image classification models that are:

*SOTA on ImageNet (86.5% top-1 w/o extra data)
*Up to 8.7x faster to train than EfficientNets to a given accuracy
*Normalizer-free (no BatchNorm!)

Paper: https://t.co/xvYDkgDCY0
Code: https://t.co/SmKU0gNCy7

[Bhav Ashok @bhavashok]

@ylecun @xamat @andrewgwils How do you think we can solve this?

[Bhav Ashok @bhavashok]

@jzlegion There was GPHIN, which played an important role in early detection of swine flu, mers, ebola, zika etc. but it was (conveniently) shut down right before the pandemic.

[Bhav Ashok @bhavashok]

Excellent summary of a rapidly growing field of research

[Jia-Bin Huang @jbhuang0604]

Neural Volume Rendering for Dynamic Scenes

NeRF has shown incredible view synthesis results, but it requires multi-view captures for STATIC scenes.

How can we achieve view synthesis for DYNAMIC scenes from a single video? Here is what I learned from several recent efforts.

[Frank Dellaert @fdellaert]

2020 was the year in which *neural volume rendering* exploded onto the scene, triggered by the impressive NeRF paper by Mildenhall et al. I wrote a post as a way of getting up to speed in a fascinating and very young field and share my journey with you: https://t.co/rGUByLBlnl

[Robert Nelsen @rtnarch]

All drug development should be like Operation Warp Speed. Alzheimer’s, Cancer. Mental health. And many more. Need this sense of urgency in government and industry all the time. Every day.

[Bhav Ashok @bhavashok]

Always be wary of inverse hormesis in health and work.

[Bhav Ashok @bhavashok]

Might be time to get that smart watch you’ve always wanted. 63% detection rate is better than chance and can only improve as the sample size increases. Not to mention the second order effects on transmission.

[Stanford Healthcare Innovation Lab @StanfordHIL]

We have been running one of the world’s most impactful research labs for 11 years.

We just published a landmark study on COVID-19.

https://t.co/EZVk5LhKiw

Here is a thread on what we learned👇👇

[Bhav Ashok @bhavashok]

Closed-loop deep learning is the ultimate antifragile sysem

[Bhav Ashok @bhavashok]

Underrated essay on the Pigeonhole principle, it concludes that: "there's a set of parents in your ancestry who are blood relatives of each other."
https://t.co/0R4ezCI8rt

[Bhav Ashok @bhavashok]

Old but great talk by @astroEdLu on preventing one of the largest existential threats to humanity
https://t.co/Ibxlvu3OA2

[priya joseph @ayirpelle]

@bhavashok Architecture Compression down to 0.2 GPU days #BayLearn2020 @BayLearnOrg

[NuronLabs @NuronLabs]

Announcing https://t.co/V7naD7OUa7 https://t.co/mbm9DLePBD

[Phys.org @physorg_com]

Scientists find upper limit for the #speedofsound @QMUL @ScienceAdvances https://t.co/SkDOQffIKy

[Bhav Ashok @bhavashok]

In general, I think that this idea of learning invariants in a domain well is a vastly unexplored one compared to the standard "generalization across many tasks/environments" narrative in ML. (6/6)

[Bhav Ashok @bhavashok]

Its amazing how much contextual creativity is unlocked when the brain 'overfits' to something. For example, advanced chess players can memorize complex positions, track entire games blindfolded, and simulate multiple turns all in their head. (1/6)

[Bhav Ashok @bhavashok]

It is interesting to note that some of these learned invariants do generalize to other chess variants such as Fischer-random or Kriegspiel chess. Deepmind's work on designing new chess variants is also interesting. https://t.co/1NGPdctykv (5/6)