andreyk

87 ms·

961.
Inside an AI Conference – Robotics Science and Systems 2018(skynettoday.com)
3 points by andreyk 8y ago|0 comments
962.
Bringing Learning to Robotics: Highlights from Robotics Science and Systems 2018(thegradient.pub)
5 points by andreyk 8y ago|0 comments
963.
by andreyk 8y ago
A good thing to read to know the basics of the game and what to pay attention to: http://smerity.com/articles/2018/n_things_to_look_out_for_in...
964.
Has AI surpassed humans at translation? No – here's why(skynettoday.com)
4 points by andreyk 8y ago|0 comments
965.
AI has a PR problem, and AI researchers should not ignore it(skynettoday.com)
3 points by andreyk 8y ago|0 comments
966.
Bringing Learning to Robotics: Highlights from Robotics Science and Systems 2018(thegradient.pub)
3 points by andreyk 8y ago|0 comments
967.
Bringing Learning to Robotics: Highlights from RSS 2018(thegradient.pub)
2 points by andreyk 8y ago|0 comments
968.
Autonomous Driving, Both Close and Far from Ubiquity(skynettoday.com)
1 points by andreyk 8y ago|0 comments
969.
by andreyk 8y ago
sure, and there has already been a ton of work on grasping (see eg DexNet 2.0, Closing the loop for robotic grasping). This stuff is already in the process of being commercialized, and this research does not advance it much.
970.
by andreyk 8y ago
sure, but they demonstrate it for a very simple use case of a single object with no background etc. ; domain randomization has yet to be demonstrated to work on more complex environments/tasks
971.
by andreyk 8y ago
I am a 2nd year Phd at Stanford and you could definitely do such work here! Also at CMU, Georgia Tech, Berkeley, U Washington, and others. You can enter PhD via EE/MechE or CS - once you are focused on research it does not matter much.
972.
by andreyk 8y ago
Not to be too negative, it's cool work, but I'd argue unlike the OpenAI result it is not so surprising this was doable with the techniques they used ; see eg this paper from Google http://www.roboticsproceedings.org&#x2
973.
by andreyk 8y ago
That is true (and I am well aware of this, as I do research on robotic grasping https://sites.google.com/view/task-oriented-grasp ). But, this would very hard to generalize to more complex manipulation where the object
974.
by andreyk 8y ago
Link to paper ( why no Arxiv :/ ): https://d4mucfpksywv.cloudfront.net/research-covers/learning... TLDR (quick-ish skim, feel free to correct) they train a deep neural network to control a robot hand to choose des
975.
by andreyk 8y ago
Basically coincidence; the models here are far simpler and do not at all mimic the human nervous system.
976.
by andreyk 8y ago
"This is a perfect example of how AI is taking over the world by storm." Because they taught a robot hand to rotate a cube well... ? Take it from an AI researcher, it's one thing to make a demo of a technique solving a narrow
977.
Show HN: Skynet This Week, a bi-weekly digest of noteworthy AI news(skynettoday.com)
2 points by andreyk 8y ago|0 comments
978.
Has AI surpassed humans at translation? No – here's why(skynettoday.com)
2 points by andreyk 8y ago|2 comments
979.
Has AI surpassed humans at translation? No – here's why(skynettoday.com)
2 points by andreyk 8y ago|0 comments
980.
Has AI surpassed humans at translation? No – here's why(skynettoday.com)
2 points by andreyk 8y ago|0 comments
981.
Has AI surpassed humans at translation? Not even close(skynettoday.com)
1 points by andreyk 8y ago|0 comments
982.
Has AI surpassed humans at translation? Not even close(skynettoday.com)
4 points by andreyk 8y ago|1 comments
983.
by andreyk 8y ago
Indeed, if these are so similar your repo really ought to be a fork of the prior repo, and you should certainly acknowledge them more in the paper (make clearer what your contributions are).
984.
David Lynch's Thoughts on Meditation, Art, and Life(youtube.com)
1 points by andreyk 8y ago|0 comments
985.
by andreyk 8y ago
True, not all properties are there to the same extent as in eg Go, but they are there to some extent and more broadly Starcraft and Dota still 'game-like' ie not as complex as driving a self driving car in the real world. The Venn
986.
by andreyk 8y ago
Yep, thanks, the idea was to highlight all the research going on and argue it deserves more attention.
987.
by andreyk 8y ago
I think AI winter is unlikely to come about this time since non-RL stuff (supervised learning) has been so successful and useful.
988.
by andreyk 8y ago
All that stuff is in part two! https://thegradient.pub/how-to-fix-rl/ "In part two, we’ll overview the different approaches within AI that can address those limitations (chiefly, meta-learning and zero-shot learni
989.
by andreyk 8y ago
To be fair, in the article itself we wind up criticizing "Pure RL" (defined as the basic formulation that is typically followed, in which all learning is done from just the reward signal) and not RL as a whole. We call out a lot o
990.
by andreyk 8y ago
yep, we in fact link to this work... "Even 5 years later, no pure RL algorithms have cracked reasoning and memory games; on the contrary, approaches that have done well at them have either used instructions <link> or demonstratio

More ›