U tweetove putem weba ili aplikacija drugih proizvođača možete dodati podatke o lokaciji, kao što su grad ili točna lokacija. Povijest lokacija tweetova uvijek možete izbrisati. Saznajte više
Can a neuroscientist understand a virtual rodent?
Authors took a realistic 3D animal, trained deep RL to control it during tasks, and then used neuroscience techniques to peak inside this animal...
[Josh Merel, Diego Aldorando, Greg Wayne, @BOlveczky]
https://arxiv.org/abs/1911.09451 pic.twitter.com/yHSro4KcvV
You can find that it learns discrete behaviors, like galloping, climbing, and stumbling! You can find how neural activity relates to choices You can find swirling rotational dynamics in the underlying 'motor' manifoldspic.twitter.com/A0HplsOaY2
this paper looks super cool, can't wait to dig into it. How do we analyze neural networks? How do we analyze behaviors? Are our neuroethological tools enough, are they accurate? Is artificial life the future?
more important question: where can I download my pet virtual rodent
sorry how could I forget to include this video of the wacky virtual rodent playing aroundpic.twitter.com/BYUSdptDR8
For the benefit of the scientific community, please enjoy these hastily cropped virtual rodents:pic.twitter.com/IHL5d8eovR
Thanks Adam! This is a great summary of what we have done so far. All credit goes to Josh, Diego and Jesse, and the computational resources at Google's DeepMind. Hopefully, much more to come from this collaboration, so stay tuned.
Thanks Adam! Hope all is well. Looking forward to hearing your thoughts!
Really liked that the virtual rodent was capable of performing a variety of complex locomotor behaviours to solve multiple tasks using a single policy. e.g., IMPALA-style setup for AC DeepRL. Since Mammals use the same general neural network to solve wide variety of tasks.
@gregeganSF we're approaching Permutation City
Twitter je možda preopterećen ili ima kratkotrajnih poteškoća u radu. Pokušajte ponovno ili potražite dodatne informacije u odjeljku Status Twittera.
