Laureline Logiaco
@laurelinelogiaco.bsky.social
Computational/theoretical neuroscientist; asst. prof. at CU Anschutz (July 2025): https://www.necolelab.com/members/laureline-logiaco.html
Particular interest in how mammalian brain regions specialize and synergize to generate hierarchical behaviors.
Particular interest in how mammalian brain regions specialize and synergize to generate hierarchical behaviors.
Reposted by Laureline Logiaco
🔥🔥🔥from @ulisespereirao.bsky.social ^^^
November 23, 2025 at 1:39 PM
🔥🔥🔥from @ulisespereirao.bsky.social ^^^
Congratulations! Really nice to see this wonderful work in final form!
October 29, 2025 at 5:14 PM
Congratulations! Really nice to see this wonderful work in final form!
As you might expect :), I definitely agree that policy gradient gets you far, and that modern robotics is basically using different flavors of PG. But progress in robotics has been a lot slower/less robust than for other fields of ML, so I am open to the idea that other 'ingredients will also matter
October 21, 2025 at 4:01 PM
As you might expect :), I definitely agree that policy gradient gets you far, and that modern robotics is basically using different flavors of PG. But progress in robotics has been a lot slower/less robust than for other fields of ML, so I am open to the idea that other 'ingredients will also matter
Cool paper & exposition of policy gradient! Nice to clarify that PG drives mean action closer to the rewarded one & reduces variance of actions, w/ parallels to motor learning. Open question if on its own PG can overcome limits: definition of reward, time, realistic output/input, non-stationarity...
October 20, 2025 at 4:38 PM
Cool paper & exposition of policy gradient! Nice to clarify that PG drives mean action closer to the rewarded one & reduces variance of actions, w/ parallels to motor learning. Open question if on its own PG can overcome limits: definition of reward, time, realistic output/input, non-stationarity...
Congratulations!
October 7, 2025 at 6:32 PM
Congratulations!
Thank you, Sean!
September 28, 2025 at 4:52 PM
Thank you, Sean!