Reposted
“The human prefrontal cortex alone contains enough neuronal hardware to run 5000 [fruit]flies.”
Meanwhile mine can’t even keep me from shitposting on Bluesky after a couple of beers.
Cool paper from @mameister4.bsky.social
authors.elsevier.com/a/1kHVa3BtfH...
Meanwhile mine can’t even keep me from shitposting on Bluesky after a couple of beers.
Cool paper from @mameister4.bsky.social
authors.elsevier.com/a/1kHVa3BtfH...
December 19, 2024 at 12:18 AM
“The human prefrontal cortex alone contains enough neuronal hardware to run 5000 [fruit]flies.”
Meanwhile mine can’t even keep me from shitposting on Bluesky after a couple of beers.
Cool paper from @mameister4.bsky.social
authors.elsevier.com/a/1kHVa3BtfH...
Meanwhile mine can’t even keep me from shitposting on Bluesky after a couple of beers.
Cool paper from @mameister4.bsky.social
authors.elsevier.com/a/1kHVa3BtfH...
Reposted
I have updated my course notes on Optimal Transport with a new Chapter 9 on Wasserstein flows. It includes 3 illustrative applications: training a 2-layer MLP, deep transformers, and flow-matching generative models. You can access it here: mathematical-tours.github.io/book-sources...
December 4, 2024 at 8:11 AM
I have updated my course notes on Optimal Transport with a new Chapter 9 on Wasserstein flows. It includes 3 illustrative applications: training a 2-layer MLP, deep transformers, and flow-matching generative models. You can access it here: mathematical-tours.github.io/book-sources...
Reposted
How to drive your research forward?
“I tested the idea we discussed last time. Here are some results. It does not work. (… awkward silence)”
Such conversations happen so many times when meetings with students. How do we move forward?
You need …
“I tested the idea we discussed last time. Here are some results. It does not work. (… awkward silence)”
Such conversations happen so many times when meetings with students. How do we move forward?
You need …
December 1, 2024 at 10:09 PM
How to drive your research forward?
“I tested the idea we discussed last time. Here are some results. It does not work. (… awkward silence)”
Such conversations happen so many times when meetings with students. How do we move forward?
You need …
“I tested the idea we discussed last time. Here are some results. It does not work. (… awkward silence)”
Such conversations happen so many times when meetings with students. How do we move forward?
You need …
Reposted
Really cool new work out of Deep Mind for video game world generation using latent diffusion! Soon you'll be able to speed run a game just by tricking a model to morph you from one location to another.
deepmind.google/discover/blo...
deepmind.google/discover/blo...
Genie 2: A large-scale foundation world model
Generating unlimited diverse training environments for future general agents
deepmind.google
December 4, 2024 at 4:31 PM
Really cool new work out of Deep Mind for video game world generation using latent diffusion! Soon you'll be able to speed run a game just by tricking a model to morph you from one location to another.
deepmind.google/discover/blo...
deepmind.google/discover/blo...
Reposted
Streaming Deep Reinforcement Learning Finally Works, by
M. Elsayed, G. Vasan, A. R. Mahmood, is one of those papers I wish I had written 😅
This paper seems to allow us to do RL with NNs as it should have always been done. Everyone should read it!
arxiv.org/abs/2410.14606
M. Elsayed, G. Vasan, A. R. Mahmood, is one of those papers I wish I had written 😅
This paper seems to allow us to do RL with NNs as it should have always been done. Everyone should read it!
arxiv.org/abs/2410.14606
Streaming Deep Reinforcement Learning Finally Works
Natural intelligence processes experience as a continuous stream, sensing, acting, and learning moment-by-moment in real time. Streaming learning, the modus operandi of classic reinforcement learning ...
arxiv.org
November 27, 2024 at 11:09 PM
Streaming Deep Reinforcement Learning Finally Works, by
M. Elsayed, G. Vasan, A. R. Mahmood, is one of those papers I wish I had written 😅
This paper seems to allow us to do RL with NNs as it should have always been done. Everyone should read it!
arxiv.org/abs/2410.14606
M. Elsayed, G. Vasan, A. R. Mahmood, is one of those papers I wish I had written 😅
This paper seems to allow us to do RL with NNs as it should have always been done. Everyone should read it!
arxiv.org/abs/2410.14606
Reposted
What are some equations or modeling approaches that are from different communities and are applied to different phenomena but you find them similar, the same, or very similar?
Example: a paper showed how replicator dynamics in evolutionary models are somewhat isomorphic with Bayes equations.
Example: a paper showed how replicator dynamics in evolutionary models are somewhat isomorphic with Bayes equations.
November 26, 2024 at 11:03 PM
What are some equations or modeling approaches that are from different communities and are applied to different phenomena but you find them similar, the same, or very similar?
Example: a paper showed how replicator dynamics in evolutionary models are somewhat isomorphic with Bayes equations.
Example: a paper showed how replicator dynamics in evolutionary models are somewhat isomorphic with Bayes equations.
Reposted
Bayes and Darwin: how replicator populations implement Bayesian computation
onlinelibrary.wiley.com/doi/full/10....
onlinelibrary.wiley.com/doi/full/10....
Bayes and Darwin: How replicator populations implement Bayesian computations
Bayesian statistical models form a cornerstone in our understanding of intelligence. As we discuss in this paper, they can be implemented by Darwinian replicator populations – pointing towards a unif...
onlinelibrary.wiley.com
November 26, 2024 at 11:29 PM
Bayes and Darwin: how replicator populations implement Bayesian computation
onlinelibrary.wiley.com/doi/full/10....
onlinelibrary.wiley.com/doi/full/10....
Reposted
I'm going to try using BlueSky more reliably for a while. Here are a few thoughts that are guiding my engagement here, and hopefully learning from our collective experience over at Twitter.
chrisholdgraf.com/blog/2024/bl...
chrisholdgraf.com/blog/2024/bl...
Thoughts on moving from Twitter/X to BlueSky - Welcome
Some quick thoughts on moving from Twitter/X to BlueSky and how I'll try to use social media after being burned once by Twitter.
chrisholdgraf.com
November 22, 2024 at 6:32 PM
I'm going to try using BlueSky more reliably for a while. Here are a few thoughts that are guiding my engagement here, and hopefully learning from our collective experience over at Twitter.
chrisholdgraf.com/blog/2024/bl...
chrisholdgraf.com/blog/2024/bl...
Reposted
Nuggets of generative diffusion
November 22, 2024 at 9:30 AM
Nuggets of generative diffusion
Reposted
There’s a single formula that makes all of your diffusion models possible: Tweedie's
Say 𝐱 is a noisy version of 𝐮 with 𝐞 ∼ 𝒩(𝟎, σ² 𝐈)
𝐱 = 𝐮 + 𝐞
MMSE estimate of 𝐮 is 𝔼[𝐮|𝐱] & would seem to require P(𝐮|𝐱). Yet Tweedie says P(𝐱) is all you need
1/3
Say 𝐱 is a noisy version of 𝐮 with 𝐞 ∼ 𝒩(𝟎, σ² 𝐈)
𝐱 = 𝐮 + 𝐞
MMSE estimate of 𝐮 is 𝔼[𝐮|𝐱] & would seem to require P(𝐮|𝐱). Yet Tweedie says P(𝐱) is all you need
1/3
November 22, 2024 at 9:45 PM
There’s a single formula that makes all of your diffusion models possible: Tweedie's
Say 𝐱 is a noisy version of 𝐮 with 𝐞 ∼ 𝒩(𝟎, σ² 𝐈)
𝐱 = 𝐮 + 𝐞
MMSE estimate of 𝐮 is 𝔼[𝐮|𝐱] & would seem to require P(𝐮|𝐱). Yet Tweedie says P(𝐱) is all you need
1/3
Say 𝐱 is a noisy version of 𝐮 with 𝐞 ∼ 𝒩(𝟎, σ² 𝐈)
𝐱 = 𝐮 + 𝐞
MMSE estimate of 𝐮 is 𝔼[𝐮|𝐱] & would seem to require P(𝐮|𝐱). Yet Tweedie says P(𝐱) is all you need
1/3