Matt McDonnell
@mattmcd
Followers
697
Following
151
Media
417
Statuses
27K
Geek Code: GS/E d s: a+ C(++++) U W++ N+ w M PS+ PE Y+ X+ !tv b+++ DI+ D e++++ r++ y?
UK
Joined June 2007
My "Complete Guide to Spherical Equivariant Graph Transformers" now has a permanent home on arXiv! 📖 📃 arXiv: https://t.co/kiGttKTiNC Over a year ago, I published this article on my blog, Alchemy Bio 🔮 ( https://t.co/2U1LtPXFpC). Even now, I’m always surprised by how often
16
103
766
Experience speeds up to 400+ Mbps to enjoy 4K streaming on multiple devices at once, working from home effectively, online gaming, social media browsing, and more. Order online in under 2 minutes.
391
2K
7K
Hockey goaltending stats are noisy. We used Bayesian spatial models to separate skill from luck, creating a fairer framework for scouting. See how Gaussian Processes clarify performance and improve decision-making. 👉 https://t.co/pHSfHcZvSz
#SportsAnalytics #PyMC #Bayesian
pymc-labs.com
Discover how Bayesian spatial modeling provides a more accurate and nuanced evaluation of hockey goaltender performance. Learn how spatial shot data and probabilistic methods reveal deeper insights...
1
14
101
Back in 2019, ARC 1 had one goal: to focus the attention of AI researchers towards the biggest bottleneck on the way to generality, the ability to adapt to novelty on the fly, which was entirely missing from the legacy deep learning paradigm. Six years later, the field has
A year ago, we verified a preview of an unreleased version of @OpenAI o3 (High) that scored 88% on ARC-AGI-1 at est. $4.5k/task Today, we’ve verified a new GPT-5.2 Pro (X-High) SOTA score of 90.5% at $11.64/task This represents a ~390X efficiency improvement in one year
42
136
2K
How do you build an end-to-end #MMM leaders can trust? Our new case study shows a transparent, reproducible #Bayesian workflow with uncertainty, lift, forecasting, and budget simulation. 👉 https://t.co/OF5ys7do37
#MarketingMixModeling #PyMCLabs
1
2
5
Topological Data Analysis does seem quite powerful, and Giotto TDA is a nice python library to try it out https://t.co/6wF6pfcynD Michael Robinson's 'Topological Signal Processing' is another great resource which focuses on sheaves for data fusion in TDA
Really getting obsessed with this subject. Many concepts being unified.
0
0
0
My favorite books to self study pure math!! When I was in undergrad, I spent a lot of time self-studying pure math. But I wasted a lot of time because I didn't have a roadmap. So here's a list of my favorite books, videos, and problem sets that you can use to self-study many
11
89
858
This paper really is groundbreaking. It solves a long-standing embarrassment in machine learning: despite all the hype around deep learning, traditional tree-based methods (XGBoost, CatBoost, random forests, etc) have dominated tabular data—the most common data format in
76
404
3K
I wrote a free online appendix on State-Space Models (SSMs) for my new book, including LMU, HiPPO, S4, and Mamba: https://t.co/sc16cX6Ggj I hope you'll find it useful! 😊
7
67
427
Can Transformers Do Everything, and Undo It Too? Check out my blog on whether language models are surjective, injective, or invertible! https://t.co/9v0gd2962J
17
98
701
Implemented Olmo 3 from scratch (in a standalone notebook) this weekend! If you are a coder, probably the best way to read the architecture details at a glance: https://t.co/wF8PkoDuBe
Olmo models are always a highlight due to them being fully transparent and their nice, detailed technical reports. I am sure I'll talk more about the interesting training-related aspects from that 100-pager in the upcoming days and weeks. In the meantime, here's the side-by-side
17
297
2K
Good survey on manifold learning. Open access. “Manifold Learning: What, How, and Why”
7
90
595
Google just dropped "Attention is all you need (V2)" This paper could solve AI's biggest problem: Catastrophic forgetting. When AI models learn something new, they tend to forget what they previously learned. Humans don't work this way, and now Google Research has a solution.
260
1K
6K
This morning I started putting together a long-ish note on the Transfer Coefficient, so I started to write about the Information Coefficient, and that sent me on a detour and I haven't even started on the TC yet. But these three pages are probably the shortest zero-to-one intro
11
27
391
@tensorqt This survey by @rballeba and collaborators seems relevant: https://t.co/nnfFW09JTw It looks like there's a line of work by Corneau, Ballester, and others, that does something along the lines you're suggesting. I'm sure @rballeba would know more!
arxiv.org
This survey provides a comprehensive exploration of applications of Topological Data Analysis (TDA) within neural network analysis. Using TDA tools such as persistent homology and Mapper, we delve...
1
2
4
my entire collection of dynamical systems animations (60fps video, no voiceovers) are publicly available for download/use in all academic contexts. go here: https://t.co/g7baJVvAr8
youtube.com
This is a series of short audio-free video simulations of dynamical systems concepts and examples. This is meant for teachers or students of dynamical system...
@robertghrist Dear @robertghrist, this is awesome and so timely...Tomorrow I'll talk about the Sitnikov problem at the Chalk and Cheese session at ETHZ as an example of a shift. Can I borrow this amazing animation?
3
5
28
taught my students how to analyze chaotic dynamics via the geometric lorenz attractor this week... such a beautiful branched surface...
28
113
913
new preprint: arithmetic barcodes for network sheaves over p-adic integers 🔗 https://t.co/ocI5Nd1uRQ persistent homology filters by geometric scale. we filter by algebraic precision instead: ℤ_p ⊇ pℤ_p ⊇ p²ℤ_p ⊇ ⋯ torsion summands ℤ_p/p^a become bars of length a,
arxiv.org
Persistent homology tracks topological features across geometric scales, encoding birth and death of cycles as barcodes. We develop a complementary theory where the filtration parameter is...
2
6
39
1/🧵 Out today in @FrontiersIn @FrontScience: “Consciousness science: where are we, where are we going, and what if we get there?" By @axelcleeremans, @Liad_Mudrik, & me (a @CIFAR_News & @ERC_Research production🧠). https://t.co/HNzCaDPsF8
43
152
550
my love-letter to the humanities: a book on mathematics & poetry. it starts with the question: "Where the Hell is Heaven?" thesis: in imagining the afterlife, poets from ancient times used the same imaginative skills that geometers and topologists use to imagine abstract
6
36
257
25 years ago, I struggled with whether to do mathematics or music as a profession. Then one of my University teachers mentioned off-hand that he'd had a similar struggle, and chose as profession what would be harder to make a hobby. Now he's written a book only he could write.
my love-letter to the humanities: a book on mathematics & poetry. it starts with the question: "Where the Hell is Heaven?" thesis: in imagining the afterlife, poets from ancient times used the same imaginative skills that geometers and topologists use to imagine abstract
13
27
316