I will be presenting on our #ICML23 paper "Polarity is all you need to learn and transfer faster"
Find me at poster #613 in session 1 (7.25 11AM)!
Let's chat about Neuro-AI Synergy, learning efficiency, and ML in general. Looking forward to connect with everyone!
How does our bio-inspired method compare with finetuning? It learns more efficiently, especially in low data regime! (red dominates purple) This result suggests polarity may be a more effective and compressed medium of network knowledge.
Why do brains have inhibitory connections? Why do deep networks have negative weights? Our answer is: to learn more functions.
Find out more from the tweet thread or pre-print https://t.co/m4sNpa736p I'll also be presenting this (plus more!) at SfN (#NS22) nanosymposium Nov. 15!
Getting the poster session ready with @Alice72666511 and @MChanCortes for the first @HopkinsNeuro recruitment of the year! SO excited to meet you all on Monday! Let us know if we can answer any and all questions about our incredible program and Baltimore 🥳
Three fantastic paintings describing parts of the science regime I’ve explored/am exploring! (Only missing oligodendrocytes) Thank you @artologica for the amazing art! #Artscape#Baltimore
Postdoctoral opportunities in hippocampal behavioral physiology. Please see
https://t.co/trdkVvbuAI for application instructions. We use multiple techniques, including electrophysiology and imaging, collaborating with theoreticians and bioengineers.
@JakeLeyhr light-sheet imaging on both brains of this 2-headed #Zebrafish. z-projection. virtual reality system, both brains are trying to take control of the spinal cord :) @MishaAhrens