RT @rodneyabrooks@twitter.com

A short blog post today, on how pull backs on predictions for autonomous cars should inform our evaluations of predictions on timing of AGI. rodneybrooks.com/agi-has-been-

๐Ÿฆ๐Ÿ”—: twitter.com/rodneyabrooks/stat

RT @RanjayKrishna@twitter.com

Announcing the first ๐—œ๐—–๐—–๐—ฉ ๐˜„๐—ผ๐—ฟ๐—ธ๐˜€๐—ต๐—ผ๐—ฝ ๐—ผ๐—ป ๐—ฆ๐—ฐ๐—ฒ๐—ป๐—ฒ ๐—š๐—ฟ๐—ฎ๐—ฝ๐—ต ๐—ฅ๐—ฒ๐—ฝ๐—ฟ๐—ฒ๐˜€๐—ฒ๐—ป๐˜๐—ฎ๐˜๐—ถ๐—ผ๐—ป ๐—ฎ๐—ป๐—ฑ ๐—Ÿ๐—ฒ๐—ฎ๐—ฟ๐—ป๐—ถ๐—ป๐—ด. If your research involves structured data or graph-based learning, consider submitting to us by August 15, 2019: sgrl.stanford.edu

๐Ÿฆ๐Ÿ”—: twitter.com/RanjayKrishna/stat

RT @Miles_Brundage@twitter.com

"Human Visual Understanding for Cognition and Manipulation -- A primer for the roboticist," Martin Hjelm: arxiv.org/abs/1905.05272

๐Ÿฆ๐Ÿ”—: twitter.com/Miles_Brundage/sta

Fun typo :P

File "scripts/train.py", line 445, in <module>
original_conf = copy.deepcopy(copy)
File "/home/andrey/Documents/envs/rl-with-experts/lib/python3.5/copy.py", line 182, in deepcopy
...
RecursionError: maximum recursion depth exceeded while calling a Python object

RT @ankurhandos@twitter.com

MeshCNN: A Network with an Edge ranahanocka.github.io/MeshCNN/

After point clouds (PoinNet style) this is an extension of neural networks to work directly on 3D meshes with mesh convolutions, pooling and unpooling operators.

๐Ÿฆ๐Ÿ”—: twitter.com/ankurhandos/status

RT @hardmaru@twitter.com

An interactive article explaining why weight initialization is so important for training neural nets by @deeplearningai_@twitter.com, written in the distill.pub format. deeplearning.ai/ai-notes/initi

๐Ÿฆ๐Ÿ”—: twitter.com/hardmaru/status/11

๐Ÿ˜‚๐Ÿ˜…๐Ÿ˜๐Ÿค”

RT @DevilleSy@twitter.com

The academic job market

๐Ÿฆ๐Ÿ”—: twitter.com/DevilleSy/status/1

RT @gradientpub@twitter.com

Difficult problems in the real world often require thinking more than one step in advance. Researchers from @IBMResearch@twitter.com discuss how we can teach the human concept of "planning" to AI.

thegradient.pub/when-ai-plans-

๐Ÿฆ๐Ÿ”—: twitter.com/gradientpub/status

RT @LMescheder@twitter.com

Check out our new blog in which we highlight our latest results. The first entry is on our CVPR 2019 paper "Occupancy Networks - Learning 3D Reconstruction in Function Space" in which propose a new output representation for 3D deep-learning. autonomousvision.github.io/occ

๐Ÿฆ๐Ÿ”—: twitter.com/LMescheder/status/

RT @wgussml@twitter.com

Excited to announce our competition: The MineRL Competition for Sample-Efficient Reinforcement Learning! With @rsalakhu@twitter.com @katjahofmann@twitter.com @diego_pliebana@twitter.com @flippnflops@twitter.com @svlevine@twitter.com @OriolVinyalsML@twitter.com @chelseabfinn@twitter.com and others!

Participate here! minerl.io/competition

๐Ÿฆ๐Ÿ”—: twitter.com/wgussml/status/112

RT @sleepinyourhat@twitter.com

This is clearly the right way to present related work.

๐Ÿฆ๐Ÿ”—: twitter.com/sleepinyourhat/sta

RT @kaifulee@twitter.com

If there are a million Tesla robo-taxis functioning on the road in 2020, I will eat them. Perhaps @rodneyabrooks@twitter.com will eat half with me? twitter.com/rodneyabrooks/stat

๐Ÿฆ๐Ÿ”—: twitter.com/kaifulee/status/11

RT @realSharonZhou@twitter.com

Officially presenting HYPE at @iclr2019@twitter.com! If you train generative models and would like to see how yours fairs against the state of the art, please see our competition at hype.stanford.edu/

If you're at the conference, come get HYPE'd today in room R02!

๐Ÿฆ๐Ÿ”—: twitter.com/realSharonZhou/sta

Supet cool finding!

"Overall, attaining models that are robust and interpretable will require explicitly encoding human priors into the training process."

RT @evolvingstuff@twitter.com

Adversarial Examples Are Not Bugs, They Are Features

"adversarial examples can be directly attributed to the presence of non-robust features ... patterns in the data distribution that are highly predictive, yet brittle and incomprehensible to humans"

arxiv.org/abs/1905.02175

๐Ÿฆ๐Ÿ”—: twitter.com/evolvingstuff/stat

RT @skynet_today@twitter.com

Our digest of the last week in AI news is out!

Click on for a summary of the major stories about AI related to advances & business, concerns & hype, analysis & policy, and more.

skynettoday.com/digests/the-tw

๐Ÿฆ๐Ÿ”—: twitter.com/skynet_today/statu

Something nice: the great video game critic @Campster@twitter.com (re-posted) a video of him trying out something new which he thought "didn't work", and the comments are FULL of positivity as well as real constructive feedback. The internet ain't so bad sometimes...
youtube.com/watch?v=reIdE6qO-Y

Interesting thought: Twitter (and other social media) enable more widespread and distributed , as opposed to a few high profile scientists doing most of the communication to broader public. Surely a good thing!

RT @julianjon@twitter.com

If you have a problem w scientists on Twitter, then don't follow them. A few reasons why I love Twitter: (1) we show scientists are ppl too, (2) learn sci outside my own discipline, (3) share gr8 work from colleagues, and (4) my mom isn't on here yet. twitter.com/Alex_Washburne/sta

๐Ÿฆ๐Ÿ”—: twitter.com/julianjon/status/1

Show more
Scholar Social

Scholar Social is a microblogging platform for researchers, grad students, librarians, archivists, undergrads, academically inclined high schoolers, educators of all levels, journal editors, research assistants, professors, administratorsโ€”anyone involved in academia who is willing to engage with others respectfully. Read more ...