Kameron Decker Harris @kamdh.bsky.social · 16hMathematicians have been writing in a language that's easier for machines to understand than humans #hottake 020
Kameron Decker Harris @kamdh.bsky.social · 18hAs a group of people to piss off, math nerds are politically quite safe. The average human seems to despise math, in my experience! 010
Kameron Decker Harris @kamdh.bsky.social · 20hSmall wrinkle: There's probably a bit of back-and-forth where evolution has changed the way bodies/brains are built to approach optimal algorithms. It's not necessarily one-way. 200
Kameron Decker Harris @kamdh.bsky.social · 20hI enjoyed Rockmore's essay on AI & math, especially the points in the end about dehumanizing obsession with speed #slowscience www.nybooks.com/online/2026/...nybooks.comThe Work of Math in the Age of Artificial Reproduction | Dan RockmoreThere are seven questions in pure mathematics deemed important enough by the Clay Mathematics Institute, a nonprofit foundation devoted to “furthering the 051
Reposted by Kameron Decker HarrisJan van Gemert @jvgemert.bsky.social · 06/10/2026In addition to the research findings, this is the type of research I really enjoyed doing 😊 "On Hyperparameter Tuning on the Test Set" arxiv.org/abs/2610.05902arxiv.orgOn Hyperparameter Tuning on the Test Set"Don't tune hyperparameters on the test set" is often stated in machine learning textbooks. Violating it is considered a cardinal sin that produces misleadingly optimistic results, corrupts benchmark ... 2224
Kameron Decker Harris @kamdh.bsky.social · 05/10/2026I let my New Yorker subscription lapse and am interested in subscribing to another magazine with excellent long- and short-form writing about all kinds of things. Any recommendations? 200
Kameron Decker Harris @kamdh.bsky.social · 05/10/2026If you think AI is going to figure out all the research you were hoping to do, maybe pivot to an area you are genuinely interested in and can make a contribution! There are so many interesting problems 020
Kameron Decker Harris @kamdh.bsky.social · 04/10/2026Bummer man, but I know how that is. I have still found it satisfying to work in smaller, cheaper ways so long as I can find smart people to talk to and think alongside. 110
Kameron Decker Harris @kamdh.bsky.social · 04/10/2026I think the point I was trying to make is: Building systems to be intelligent or solve specific problems is generally agnostic to how biological systems "do" those things. Actually figuring out why bodies break down or mental states go awry depends on all the wet murky details 150
Kameron Decker Harris @kamdh.bsky.social · 04/10/2026Maybe you're too embedded in the ai hype space, if you think it's going to quickly "solve" all these things 130
Kameron Decker Harris @kamdh.bsky.social · 04/10/2026Also: the brain is about so much more than intelligence narrowly scoped (as in AI) because of health, biology, physics, etc etc 130
Kameron Decker Harris @kamdh.bsky.social · 04/10/2026Imo, there's a lot that's still unknown in both brains AND artificial networks that keeps me interested and engaged and not feeling discouraged. Maybe its because I haven't felt the need to stay abreast of all the ML/DL/AI progress? 120
Kameron Decker Harris @kamdh.bsky.social · 04/10/2026I don't think it's really brute force but something more delicate. And it isn't clear yet what the functional spaces brains use are (again, they're different than machines) 000
Kameron Decker Harris @kamdh.bsky.social · 04/10/2026I think this summary is sort of right and sort of wrong: yes current AI paradigm is solving a lot of intelligence problems. But it's producing an alien intelligence that works quite differently than animals 140
Kameron Decker Harris @kamdh.bsky.social · 03/10/2026This is hilarious. I kinda wonder if that account is a bot but I don't think so 010
Reposted by Kameron Decker HarrisMattia Rigotti @matrig.net · 03/10/2026The pope just got... schmidhuber'd 1183
Kameron Decker Harris @kamdh.bsky.social · 03/10/2026Yeah the organizing my disorganized class materials use case is very nice 011
Reposted by Kameron Decker HarrisDurstewitzLab @durstewitzlab.bsky.social · 01/10/2026In our #NeurIPS2026 spotlight “Parallel-in-Time Training of Recurrent Neural Networks for Dynamical Systems (DS) Reconstruction” (preprint: arxiv.org/abs/2605.12683) we speed up training of nonlinear RNNs on time series from chaotic DS by >100x by combining DEER with generalized teacher forcing. 1154
Kameron Decker Harris @kamdh.bsky.social · 02/10/2026Lorenz 63 is like the MNIST of nonlinear dynamics... And I mean it in a good way! 010
Kameron Decker Harris @kamdh.bsky.social · 30/09/2026Thick and crispy, tastes of cardboard. Will dip in tea 000
Reposted by Kameron Decker HarrisCarl Zimmer @carlzimmer.com · 27/09/2026Four days ago, a scientist was surprised by the news that Anthropic discovered new virus genes for making DNA. He says he’s been studying them for years—and feeding his data to Anthropic’s AI as part of his own research. Coincidence? Here’s my story. nyti.ms/3VSWaycnyti.msDid Anthropic’s A.I. Really Make a Scientific Discovery on Its Own? (Gift Article)An expert at the University of Copenhagen said his team had been sharing its research with the company’s A.I. model, Claude, and that its new finding matched their work. 882124874
Kameron Decker Harris @kamdh.bsky.social · 28/09/2026SIAM MDS '26 registration fees: $625 (member), $425 (1-day), $845 (non-member) 🤦 www.siam.org/conferences-...siam.orgMDS26 Registration | SIAMRegistration information for the SIAM Conference on Mathematics of Data Science (MDS26) 000
Kameron Decker Harris @kamdh.bsky.social · 26/09/2026Last visit to Allen Inst we were given a tour of open scope by Severine Durand who told us about this wild work 030
Kameron Decker Harris @kamdh.bsky.social · 26/09/2026My hunch for a while is that it's a bit of both structured and random 010
Kameron Decker Harris @kamdh.bsky.social · 26/09/2026Cool work! Despite being someone who has worked on random models, it makes sense there would also be structure. An interesting question for me (which would require connectomics for multiple animals) is whether there's still variability for this and if the structure shown here is just "on average." 120
Kameron Decker Harris @kamdh.bsky.social · 26/09/2026I prefer to bash my head against mathematics until it fits inside 110
Kameron Decker Harris @kamdh.bsky.social · 26/09/2026Cool! Seems like it could be a neat class project for deep learning. Were these private data? 010
Kameron Decker Harris @kamdh.bsky.social · 24/09/2026I think the problems that they're working on are really cool, though. 000
Kameron Decker Harris @kamdh.bsky.social · 24/09/2026To me, this reads as a clear example of privatization of basic science. If the private companies are the ones to make these discoveries, the rest of us won't get to see the benefits. 200
Kameron Decker Harris @kamdh.bsky.social · 24/09/2026I just saw a clip of it. I agree with the take that doomerism and machine god thought is harmful 120
Kameron Decker Harris @kamdh.bsky.social · 23/09/2026Nice piece, thanks for the gift article. I appreciate the compressed sensing history interwoven. I'm still working in that area! 130
Reposted by Kameron Decker HarrisAnn Kennedy @antihebbiann.bsky.social · 22/09/2026Preprint time! This is a cool one. Complex systems, be they neural nets, interacting genes, or power grids, can produce diverse dynamics-- things like point attractors, limit cycles, and chaos. The dynamic landscape you get depends on how elements interact. But those interactions can also change! 28125
Kameron Decker Harris @kamdh.bsky.social · 20/09/2026A clip from the show youtu.be/7hqCct8t0U4?...youtu.be"Mint Tea" Johnny Blue Skies and the Dark Clouds 9/19/26 Vancouver YouTube video by Beth Boylan 000
Kameron Decker Harris @kamdh.bsky.social · 20/09/2026Caught an incredible show by Johnny Blue Skies (the jam band version of Sturgill Simpson). Definitely check them out if you can! I went with my father in law Alan who is a longtime deadhead and retired infectious disease professor 100
Kameron Decker Harris @kamdh.bsky.social · 20/09/2026Yes. It's already being heavily used in those areas 010
Kameron Decker Harris @kamdh.bsky.social · 19/09/2026I think I just self-destructed one of my GPUs! 000
Kameron Decker Harris @kamdh.bsky.social · 18/09/2026Saw a guy with an "I ❤️ ICE" shirt at the bike trails, no less, ugh but I guess I know I don't have to be nice 010
Reposted by Kameron Decker HarrisSweta Agrawal @hokuba.bsky.social · 16/09/2026The Agrawal lab has published its first paper! "Sticking the landing: leg mechanosensory feedback coordinates the transition from flight to landing" www.biorxiv.org/content/10.6... This project was led by a technician in the lab, Tse-Wei (Wayne) Kuo. Brief 🧵on what we foundagrawal-lab-workspace.slack.comSlack 88228
Kameron Decker Harris @kamdh.bsky.social · 17/09/2026Ugh that's terrible! The work looks amazing!!! I hope Wayne will have another chance (maybe outside the US) to continue in science 010
Kameron Decker Harris @kamdh.bsky.social · 16/09/2026I recommend reading the article, especially for an entertaining detour into skateboarding. What do you think? 000
Kameron Decker Harris @kamdh.bsky.social · 16/09/2026With so much compute, LLMs can perform informed yet random search in idea space, and it seems reasonable to find new kinds of approaches. I’m not sure we’ve seen evidence for it yet, but I think it's possible. 100