Hadi Vafaii
@hadivafaii.bsky.social
Postdoc at UC Berkeley, Redwood Center | 🧠🤖 | 🎹 | 🎾 | https://mysterioustune.com/
This one is an all time favorite of mine:
direct.mit.edu/neco/article...
(tho maybe not that relevant to the "rescue" thing)
direct.mit.edu/neco/article...
(tho maybe not that relevant to the "rescue" thing)
Natural Gradient Works Efficiently in Learning
Abstract. When a parameter space has a certain underlying structure, the ordinary gradient of a function does not represent its steepest direction, but the natural gradient does. Information geometry ...
direct.mit.edu
November 24, 2025 at 4:23 PM
This one is an all time favorite of mine:
direct.mit.edu/neco/article...
(tho maybe not that relevant to the "rescue" thing)
direct.mit.edu/neco/article...
(tho maybe not that relevant to the "rescue" thing)
🔗 Meeting summary: sensorimotorai.github.io/2025/11/20/d...
📜 Read the paper: www.pnas.org/doi/10.1073/...
💬 Join the conversation on Slack: join.slack.com/t/sensorimot...
🧠🤖🧠📈
📜 Read the paper: www.pnas.org/doi/10.1073/...
💬 Join the conversation on Slack: join.slack.com/t/sensorimot...
🧠🤖🧠📈
The neuron as a direct data-driven controller (DD-DC) by Moore et al., 2024
We explored DD-DC, a provocative normative theory proposing that neurons are not just information processors, but active feedback controllers that steer their environment—including other neurons—towar...
sensorimotorai.github.io
November 21, 2025 at 6:06 AM
🔗 Meeting summary: sensorimotorai.github.io/2025/11/20/d...
📜 Read the paper: www.pnas.org/doi/10.1073/...
💬 Join the conversation on Slack: join.slack.com/t/sensorimot...
🧠🤖🧠📈
📜 Read the paper: www.pnas.org/doi/10.1073/...
💬 Join the conversation on Slack: join.slack.com/t/sensorimot...
🧠🤖🧠📈
💡 The killer insight: biological "quirks" are actually optimal control solutions:
🔷 STDP = solving the LQR cost (learning causal impact from feedback)
🔷 Noise = stabilizing "persistence of excitation"
🔷 ReLU = optimal policy to cross unstable fixed points
Conclusion:
✅ Agency is cellular.
🧠🤖🧠📈
🔷 STDP = solving the LQR cost (learning causal impact from feedback)
🔷 Noise = stabilizing "persistence of excitation"
🔷 ReLU = optimal policy to cross unstable fixed points
Conclusion:
✅ Agency is cellular.
🧠🤖🧠📈
November 21, 2025 at 6:06 AM
💡 The killer insight: biological "quirks" are actually optimal control solutions:
🔷 STDP = solving the LQR cost (learning causal impact from feedback)
🔷 Noise = stabilizing "persistence of excitation"
🔷 ReLU = optimal policy to cross unstable fixed points
Conclusion:
✅ Agency is cellular.
🧠🤖🧠📈
🔷 STDP = solving the LQR cost (learning causal impact from feedback)
🔷 Noise = stabilizing "persistence of excitation"
🔷 ReLU = optimal policy to cross unstable fixed points
Conclusion:
✅ Agency is cellular.
🧠🤖🧠📈
Elegant theoretical derivations are exclusive to physics. Right?? Wrong!
In a new preprint, we:
✅ "Derive" a spiking recurrent network from variational principles
✅ Show it does amazing things like out-of-distribution generalization
👉[1/n]🧵
w/ co-lead Dekel Galor & PI @jcbyts.bsky.social
🧠🤖🧠📈
In a new preprint, we:
✅ "Derive" a spiking recurrent network from variational principles
✅ Show it does amazing things like out-of-distribution generalization
👉[1/n]🧵
w/ co-lead Dekel Galor & PI @jcbyts.bsky.social
🧠🤖🧠📈
November 19, 2025 at 4:19 PM
Additional links:
📝 Meeting summary: sensorimotorai.github.io/2025/11/13/a...
📅 Future meetings: sensorimotorai.github.io/schedule/
💬 Join the conversation on Slack: join.slack.com/t/sensorimot...
🧠🤖🧠📈
📝 Meeting summary: sensorimotorai.github.io/2025/11/13/a...
📅 Future meetings: sensorimotorai.github.io/schedule/
💬 Join the conversation on Slack: join.slack.com/t/sensorimot...
🧠🤖🧠📈
RL Debates 5: Anne “not everything is RL” Collins
In our 5th RL Debates presentation, Anne argued that reward-based learning is not always driven by RL computations. Sometimes it’s working memory combined with outcome-insensitive habit formation, mim...
sensorimotorai.github.io
November 19, 2025 at 7:28 AM
Additional links:
📝 Meeting summary: sensorimotorai.github.io/2025/11/13/a...
📅 Future meetings: sensorimotorai.github.io/schedule/
💬 Join the conversation on Slack: join.slack.com/t/sensorimot...
🧠🤖🧠📈
📝 Meeting summary: sensorimotorai.github.io/2025/11/13/a...
📅 Future meetings: sensorimotorai.github.io/schedule/
💬 Join the conversation on Slack: join.slack.com/t/sensorimot...
🧠🤖🧠📈
Additional links:
📝 Meeting summary: sensorimotorai.github.io/2025/novembe...
📅 Future meetings: sensorimotorai.github.io/schedule/
💬 Join the conversation on Slack: join.slack.com/t/sensorimot...
🧠🤖🧠📈
📝 Meeting summary: sensorimotorai.github.io/2025/novembe...
📅 Future meetings: sensorimotorai.github.io/schedule/
💬 Join the conversation on Slack: join.slack.com/t/sensorimot...
🧠🤖🧠📈
RL Debates 4: Adam “I literally measured value in the brain” Lowet
In our 4th RL Debates presentation, Adam presented broad topics on RL and the brain, including his distributional RL paper.
sensorimotorai.github.io
November 10, 2025 at 3:37 AM
Additional links:
📝 Meeting summary: sensorimotorai.github.io/2025/novembe...
📅 Future meetings: sensorimotorai.github.io/schedule/
💬 Join the conversation on Slack: join.slack.com/t/sensorimot...
🧠🤖🧠📈
📝 Meeting summary: sensorimotorai.github.io/2025/novembe...
📅 Future meetings: sensorimotorai.github.io/schedule/
💬 Join the conversation on Slack: join.slack.com/t/sensorimot...
🧠🤖🧠📈
📅 Next meeting: November 6, 9:00 AM PT
Where Adam Lowet will present his paper on distributional RL in brain: nature.com/articles/s41...
- See the full schedule here: sensorimotorai.github.io/schedule/
- Join our Slack: join.slack.com/t/sensorimot...
🧠🤖🧠📈
Where Adam Lowet will present his paper on distributional RL in brain: nature.com/articles/s41...
- See the full schedule here: sensorimotorai.github.io/schedule/
- Join our Slack: join.slack.com/t/sensorimot...
🧠🤖🧠📈
An opponent striatal circuit for distributional reinforcement learning - Nature
D1- and D2-expressing striatal neurons encode separate parts of a learned reward distribution, paralleling modern approaches in machine learning.
nature.com
October 23, 2025 at 9:52 PM
📅 Next meeting: November 6, 9:00 AM PT
Where Adam Lowet will present his paper on distributional RL in brain: nature.com/articles/s41...
- See the full schedule here: sensorimotorai.github.io/schedule/
- Join our Slack: join.slack.com/t/sensorimot...
🧠🤖🧠📈
Where Adam Lowet will present his paper on distributional RL in brain: nature.com/articles/s41...
- See the full schedule here: sensorimotorai.github.io/schedule/
- Join our Slack: join.slack.com/t/sensorimot...
🧠🤖🧠📈
📜 Read the summary post: sensorimotorai.github.io/schedule/
@cortical-canonical.bsky.social @thousandbrains.org
@cortical-canonical.bsky.social @thousandbrains.org
Schedule
Monthly journal club and RL Debate Series
sensorimotorai.github.io
October 23, 2025 at 9:52 PM
📜 Read the summary post: sensorimotorai.github.io/schedule/
@cortical-canonical.bsky.social @thousandbrains.org
@cortical-canonical.bsky.social @thousandbrains.org
Next Thursday (Oct 23) Niels Leadholm from @thousandbrains.org will present.
- See the full schedule here: sensorimotorai.github.io/schedule/
- Join our Slack: join.slack.com/t/sensorimot...
🧠🤖🧠📈
- See the full schedule here: sensorimotorai.github.io/schedule/
- Join our Slack: join.slack.com/t/sensorimot...
🧠🤖🧠📈
Schedule
Monthly journal club and RL Debate Series
sensorimotorai.github.io
October 18, 2025 at 6:26 AM
Next Thursday (Oct 23) Niels Leadholm from @thousandbrains.org will present.
- See the full schedule here: sensorimotorai.github.io/schedule/
- Join our Slack: join.slack.com/t/sensorimot...
🧠🤖🧠📈
- See the full schedule here: sensorimotorai.github.io/schedule/
- Join our Slack: join.slack.com/t/sensorimot...
🧠🤖🧠📈
Also, fun fact: Fritz's paper (frontiersin.org/journals/neu...) talks about world modeling in 2013, way before it was cool.
You can hear at 49:01 (youtube.com/watch?v=rlF-...), he's describing an agent that runs an "internal simulation" of the world to decide its next action.
You can hear at 49:01 (youtube.com/watch?v=rlF-...), he's describing an agent that runs an "internal simulation" of the world to decide its next action.
Frontiers | Learning and exploration in action-perception loops
Discovering the structure underlying observed data is a recurring problem in machine learning with important applications in neuroscience. It is also a prima...
frontiersin.org
October 18, 2025 at 6:26 AM
Also, fun fact: Fritz's paper (frontiersin.org/journals/neu...) talks about world modeling in 2013, way before it was cool.
You can hear at 49:01 (youtube.com/watch?v=rlF-...), he's describing an agent that runs an "internal simulation" of the world to decide its next action.
You can hear at 49:01 (youtube.com/watch?v=rlF-...), he's describing an agent that runs an "internal simulation" of the world to decide its next action.
Can I also join? I still haven't written a book, but definitely planning to. I've started with blog posts, where I break down difficult technical concepts and make the accessible:
- World Models, Adaptation, and Surprise: bsky.app/profile/hadi...
- Perception as Inference: bsky.app/profile/hadi...
- World Models, Adaptation, and Surprise: bsky.app/profile/hadi...
- Perception as Inference: bsky.app/profile/hadi...
Every living system must adapt, or die.
In my first blog post, I show how this fundamental principle can be mathematized:
✅ Brains adapt, and adaptation is about KL divergence minimization.
Let's unpack the main insights 🧵[1/n]
Link: mysterioustune.com/2025/01/13/w...
🧠🤖🧠📈 #AI
In my first blog post, I show how this fundamental principle can be mathematized:
✅ Brains adapt, and adaptation is about KL divergence minimization.
Let's unpack the main insights 🧵[1/n]
Link: mysterioustune.com/2025/01/13/w...
🧠🤖🧠📈 #AI
October 16, 2025 at 6:30 PM
Can I also join? I still haven't written a book, but definitely planning to. I've started with blog posts, where I break down difficult technical concepts and make the accessible:
- World Models, Adaptation, and Surprise: bsky.app/profile/hadi...
- Perception as Inference: bsky.app/profile/hadi...
- World Models, Adaptation, and Surprise: bsky.app/profile/hadi...
- Perception as Inference: bsky.app/profile/hadi...
Join us on Slack: join.slack.com/t/sensorimot...
More info on our website: sensorimotorai.github.io
🧠🤖🧠📈
More info on our website: sensorimotorai.github.io
🧠🤖🧠📈
Join Sensorimotor AI on Slack
Slack is a new way to communicate with your team. It’s faster, better organized, and more secure than email.
join.slack.com
October 2, 2025 at 11:31 PM
Join us on Slack: join.slack.com/t/sensorimot...
More info on our website: sensorimotorai.github.io
🧠🤖🧠📈
More info on our website: sensorimotorai.github.io
🧠🤖🧠📈