Postdoc in Computational Neuroscience, PhD in Robotics from the University of Washington. Musician/composer. I work on beep boops and notey bois. He/him.
Seattle, WA
Joined February 2015
- Tweets427
- Following2.1K
- Followers416
- Likes14.2K
Jake Sacks retweeted
(1/n) How can we infer single-trial communication between 🧠 regions? Check out our ICML 2025 paper: “Accurate Identification of Communication Between Multiple Interacting Neural Populations,” by Belle Liu and @jisacks3 in my lab: arxiv.org/abs/2506.19094
Jake Sacks retweeted
I will die on this one: pay is the single largest gatekeeper we have in academia.
This crap isn’t viable.
It’s hurting people, and limiting the pool of talented folks that think it’s a good idea to do science or have the resources to do so.
Jake Sacks retweeted
The problem with the Nobel prize is not just that they've again awarded not a single woman, although they should certainly be ashamed of themselves. The problem is our deference to these fossilised patriarchal elitist institutions, reflecting broader issues in scientific culture.
Jake Sacks retweeted
[Me at age thirteen listening to Trent Reznor scream about the violent self-destruction and bottomless despair of heroin addiction] god this is just like going to middle school
Jake Sacks retweeted
Wow, what a great Summer Workshop on the Dynamic Brain!
It's hard to believe that SWDB is already celebrating 10 years. Time really flies by when you are talking cool science in the gorgeous San Juan Islands! 🧠🌲🌊
Jake Sacks retweeted
The PAMS Program application will go LIVE on September 10! Sign up for an alert to apply and be matched with a PhD mentor who will provide personalized feedback and guidance on your application journey.cs.washington.edu/academics/… @uwcse @csenews
Interested in using task- and data-trained RNNs to identify signatures of learning-related changes in upstream inputs vs. local connectivity in M1? Come to our poster [2-007] today at #cosyne24 during poster session 2 (12:30-3:45 PM) to learn more and say hi!
Joint work with Emily Oby, @jehosafet, Alan Degenhart, Patrick Sadtler, Kristin Quick, Stephen Ryu, Elizabeth Tyler-Kabara, Steven Chase, @yulikeneuro, @aaronbatista, and @MattGolub_Neuro!
Jake Sacks retweeted
2. Deep Model Predictive Optimization (Deployable Workshop Mon 10:40). DMPO learns the inner-loop optimizer of sampling-based MPC directly via experience, outperforming MPC and end-to-end RL baselines. arxiv.org/abs/2310.04590
Best graphical abstract I've ever seen
The study of plasticity has always been about gradients. We mean it. All the things we talk about are in a way dealing with important aspects of gradients. Gradient descent yes. Simple implementation no.
physoc.onlinelibrary.wiley.c…
Me after an hour of work: wow what a day, time to play Zelda till I literally collapse from exhaustion
Jake Sacks retweeted
Can we narrow the reality gap for vision-based tactile sensors? We present Tactile Diffusion for generating synthetic tactile images. On real Braille reading task with a DIGIT sensor, a classifier trained with our model outperforms other data adaptation approaches. (1/5)
Jake Sacks retweeted
This not only lures people into uncritically trusting the outputs of systems like ChatGPT, but also misattributes agency. Accountability properly lies not with the artifacts but with their builders.
What we need is regulation that enforces transparency.
Jake Sacks retweeted
ai "safety" versus ai ethics. can you spot the difference?
and here's a good thread by @Soccermatics where i got these pictures from:
xitter.cf/Soccermatics/status/16…
Jake Sacks retweeted
The main emergent phenomenon I've observed from LLMs is computer scientists on Twitter emerging as philosophers, cognitive scientists, biologists, linguists, ethicists, artists, ...
Jake Sacks retweeted
ONE-WINGED ANGEL, BUT IT GETS REALLY EMO. I took more than a few liberties with the chorus but some of it sounds sort of like the original a little bit...kinda. Go watch the full thing here: youtu.be/-TCl5xpIcYU