Simone Azeglio @NeurIPS
From Physics to Vision Neuroscience & AI | PhD Candidate @InstVisionParis & @ENS_ULM | Enjoyed my time @FlatironCCN, @CERN | co-organizer @neurreps
- 🧠🔬 Excited to share our #NeurIPS2025 paper: "Convolution Goes Higher-Order"! We asked: Can shallow networks be as expressive as deep ones? Inspired by biological vision, we introduce higher-order convolutions that capture complex image patterns standard CNNs miss. 🧵👇
- The issue: Standard CNNs use pointwise nonlinearities with tied weights. When you expand σ(w₁x₁ + w₂x₂), different orders share the same weights—limiting expressivity! Natural images have rich higher-order correlations that basic convolutions struggle to capture.
- Our solution: Higher-order convolutions! We extend standard convolution to include learnable 2nd, 3rd, and 4th order terms with independent weights. Each order captures different aspects of visual structure—from edges (1st) to complex textures (3rd/4th).
-
View full threadLink to Poster & Paper --> neurips.cc/virtual/2025...
- 🧠 How do neurons encode information? We know HOW MUCH, but what about WHAT information they encode? Our new work uses diffusion models to decompose neural information down to individual stimuli & features! 🎯Spotlight at #NeurIPS2025 🌟📄 arxiv.org/abs/2505.11309
- The challenge: decomposing mutual information I(R;X) into stimulus-specific contributions I(x) is fundamentally ill-posed - many solutions exist! Previous methods (Fisher info, stimulus specific info) violate key properties, making them hard to interpret as sensitivity measures.
- Our approach: introduce 4 core axioms any meaningful decomposition should satisfy: ✓ Completeness: recovers total mutual information ✓ Locality: local changes have local effects ✓ Positivity: information ≥ 0 ✓ Additivity: combines measurements properly
-
View full threadThis project was jointly done with @slneuro.bsky.social building on deep intuitions and under the supervision of Matthew Chalk! 🙏 Come discuss at #NeurIPS2025! 🎪 📍 Fri Dec 5, 11 AM-2 PM PST 📍 Exhibit Hall C,D,E #2005
- Reposted by Simone Azeglio @NeurIPS[Not loaded yet]
- Reposted by Simone Azeglio @NeurIPS🕳️🐇Into the Rabbit Hull – Part II Continuing our interpretation of DINOv2, the second part of our study concerns the *geometry of concepts* and the synthesis of our findings toward a new representational *phenomenology*: the Minkowski Representation Hypothesis
- Reposted by Simone Azeglio @NeurIPSFiguring out how the brain uses information from visual neurons may require new tools, writes @neurograce.bsky.social. Hear from 10 experts in the field. #neuroskyence www.thetransmitter.org/the-big-pict...
- Reposted by Simone Azeglio @NeurIPS[Not loaded yet]
- Reposted by Simone Azeglio @NeurIPSImages (or image patches) are secretly multi-channel signals over groups. Below, the dihedral group of order 8: reflecting/rotating the image permutes the values in the magenta vector. So we can reshape the image into 8-tuples that all permute according to the dihedral group (edge case diagonals).
- Reposted by Simone Azeglio @NeurIPSThrilled to see this work accepted at NeurIPS! Kudos to @hafezghm.bsky.social for the heroic effort in demonstrating the efficacy of seq-JEPA in representation learning from multiple angles. #MLSky 🧠🤖
- Reposted by Simone Azeglio @NeurIPSWe present our preprint on ViV1T, a transformer for dynamic mouse V1 response prediction. We reveal novel response properties and confirm them in vivo. With @wulfdewolf.bsky.social, Danai Katsanevaki, @arnoonken.bsky.social, @rochefortlab.bsky.social. Paper and code at the end of the thread! 🧵1/7
- Reposted by Simone Azeglio @NeurIPSOur illustrated guide to non-Euclidian ML is finally published! Check it out for ⭐️ gorgeous figures (with new additions!) on topology, algebra, and geometry in the field ⭐️ broken down tables for easy reading ⭐️ accessible text, additional refs, and more iopscience.iop.org/article/10.1...
- Reposted by Simone Azeglio @NeurIPS🚨New paper🚨 Neural manifolds went from a niche-y word to an ubiquitous term in systems neuro thanks to many interesting findings across fields. But like with any emerging term, people use it very differently. Here, we clarify our take on the term, and review key findings & challenges rdcu.be/ex8hW
- Reposted by Simone Azeglio @NeurIPSDo you study neural systems with feedback at different temporal-, spatial-, hierarchical-, data-, or computational- scales? Have you submitted your abstract to the "Neurocybernetics at Scale" symposium? Due to multiple requests, new abstract submission deadline is 18 July! #AI4Science #cybernetics
- NeurReps is back for its 4th edition at NeurIPS 2025! Stay tuned for updates!
- 📢 Call for Papers: NeurReps 2025 ‼️‼️‼️ 🧠 Submit your research on symmetry, geometry, and topology in artificial and biological neural networks. Two tracks: Proceedings (9 pages) and Extended Abstract (4 pages). Deadline: Aug 22, 2025. www.neurreps.org/call-for-pap...
- Reposted by Simone Azeglio @NeurIPSExciting new preprint from the lab: “Adopting a human developmental visual diet yields robust, shape-based AI vision”. A most wonderful case where brain inspiration massively improved AI solutions. Work with @zejinlu.bsky.social @sushrutthorat.bsky.social and Radek Cichy arxiv.org/abs/2507.03168
- Reposted by Simone Azeglio @NeurIPS[Not loaded yet]
- Reposted by Simone Azeglio @NeurIPS
- Reposted by Simone Azeglio @NeurIPS🧠✨ Join us for LAMINR: Learning and Aligning Manifolds of Single-Neuron Invariances — uncover the geometry of thought through implicit neural representations! 📅 Mon, Jun 16 @ 18:00 CET 🔗 Zoom | PW: 314159 🎙️ Part of the #NeurReps Global Seminar Series #AI #Neuroscience #RepresentationLearning
- NeurReps 2025 is starting to take shape! If you're interested in serving in the Program Committee 👇
- NeurReps 2025 is on its way ! If you're interested in serving in the Program Committee, consider filling this form ! forms.gle/wgUXHT7Mq71n...
- Reposted by Simone Azeglio @NeurIPSExcited to share our latest study! 🧠✨ We reveal how non-direction-selective retinal ganglion cells encode motion beyond their receptive field, relaying the unconventional signals to the brain. A new insight into multimodal neurons in visual processing! doi.org/10.1073/pnas...
- Reposted by Simone Azeglio @NeurIPSWhy do diverse ANNs resemble brain representations? Check out our new paper with Colton Casto, @nogazs.bsky.social , Colin Conwell, Mark Richardson, & @evfedorenko.bsky.social on “Universality of representation in biological and artificial neural networks.” 🧠🤖 tinyurl.com/yckndmjt
- Reposted by Simone Azeglio @NeurIPSLooking for the best systems #neuroscience #phd program? Champalimaud Centre for the Unknown has one! #PhDposition = fully funded. US-style lab rotations and an awesome community in Lisbon, Portugal. Apply by the end of Jan, 2025. bit.ly/4iB9SwP
- Reposted by Simone Azeglio @NeurIPSI dream of a future in which single neuron dynamics ➡️ circuit motifs ➡️ distributed computations are taught as part of the same curriculum and no one bats an eye and everyone is happy.
- Reposted by Simone Azeglio @NeurIPSNew results! Visual adaptation changes the geometry of V1 population activity: frequent stimuli elicit smaller responses but become more discriminable. Similar results are seen in ANNs trained with metabolic constraints, suggesting these changes emerge from efficient coding. bit.ly/3VJHXRn
- Reposted by Simone Azeglio @NeurIPSCreated a starter pack of neuroscience in/from Paris. Let me know if you want to be added (the 'from' can include those not in Paris anymore) or just tap in if you want to know what we're talking about! Regardless, please re-tweet! go.bsky.app/3Zs9w5wat://did:plc:mmqv5bnhnnlbbywvzdsbxkkw/app.bsky.graph.starterpack/3lczwvunk4m2u
- Reposted by Simone Azeglio @NeurIPS[Not loaded yet]
- Reposted by Simone Azeglio @NeurIPSEero Simoncelli on the geometry of the distribution of natural images
- Reposted by Simone Azeglio @NeurIPSNeurReps has begun! Join us today for a program combining geometry/topology, deep learning, neuroscience, and interpretability
- Reposted by Simone Azeglio @NeurIPSOur second Keynote speaker is @taliakonkle.bsky.social from Harvard University. Come to hear about "The Cortical Topography of Visual Space Representation"!
- Reposted by Simone Azeglio @NeurIPSJoin us for our next Keynote invited talk from Stefanie Jegelka on "In-context Symmetries via Contextual World Models"!
- Reposted by Simone Azeglio @NeurIPSEnigma's at NeurIPS - come find us. New jobs posted on our website for research scientists & engineers in multi-modal modeling & interpretability enigmaproject.ai
- Join us this Saturday @neuripsconf.bsky.social !
- Ready for #NeurReps2024? Our schedule at @neuripsconf.bsky.social is out 📅 neurips.cc/virtual/2024... Join us this Saturday! 🧠 With @eerosim.bsky.social @taliakonkle.bsky.social Stefanie Jegelka, Srinivas Turaga, Ilker Yildirim and Savannah Thais
- Reposted by Simone Azeglio @NeurIPSExcited to talk with the community about what we've been working on! Co-founding @newtheory.ai with @colin-r-odonnell.bsky.social to build new foundational architectures for intelligence. Come join us at NeurIPS in Vancouver this week for our soft-launch event 🎊🥂 partiful.com/e/rgVdt2XkzZ...
- Reposted by Simone Azeglio @NeurIPS[Not loaded yet]
- Reposted by Simone Azeglio @NeurIPSAnother cool "digital twin" work from Wood’s lab After the virtual chicks, this time: "We embodied artificial neural networks in artificial fish and raised the artificial fish in virtual fish tanks that mimicked the rearing conditions of biological fish" #NeuroAI www.nature.com/articles/s41...
- Reposted by Simone Azeglio @NeurIPSHow do interneurons reshape neural responses? I'm excited to present work with @eerosim.bsky.social at #NeurIPS2024 that proposes a nonlinear recurrent circuit model motivated by efficient coding theory. Poster: 4:30p on Fri, Dec 13 Paper: openreview.net/forum?id=ojL...
- I'll be in Vancouver for the next week @neuripsconf.bsky.social. Who will I see there? Happy to chat about vision, comp neuro, equivariance, learning representations. In case you share my interests, consider joining our workshop @neurreps.bsky.social on Dec 14th 🧠
- Reposted by Simone Azeglio @NeurIPSNew 716 hour EMG dataset just dropped ✨ ai.meta.com/blog/open-so...
- Reposted by Simone Azeglio @NeurIPS[Not loaded yet]
- Reposted by Simone Azeglio @NeurIPSThe Undergrad Travel Grant Program (deadline: Dec 6th) provides an opportunity for undergrads to learn more about comp systems neuro. It’s especially suited for students considering neuro graduate study! Apply here: docs.google.com/forms/d/e/1F... More info: www.cosyne.org/travel-grants
- Reposted by Simone Azeglio @NeurIPSExcited to release what we’ve been working on at Amaranth Foundation, our latest whitepaper, NeuroAI for AI safety! A detailed, ambitious roadmap for how neuroscience research can help build safer AI systems while accelerating both virtual neuroscience and neurotech. 1/N
- Reposted by Simone Azeglio @NeurIPS[Not loaded yet]
- Reposted by Simone Azeglio @NeurIPS🚨 We just updated our perspective on representational alignment. The most recent version is both more crisp and more comprehensive. We try to find common language across research disciplines for aligning the representations from different info processing systems! arxiv.org/abs/2310.13018
- Reposted by Simone Azeglio @NeurIPSSorbonne Université est heureuse de rejoindre officiellement BlueSky ! 🦋 Nous avons hâte de partager ici notre actualité recherche et de contribuer ainsi aux débats scientifiques et sociétaux ! 🎓 Abonnez-vous pour suivre notre actualité ! #HelloESR #ESR
- Reposted by Simone Azeglio @NeurIPSMy deep learning course at the University of Geneva is available on-line. 1000+ slides, ~20h of screen-casts. Full of examples in PyTorch. fleuret.org/dlc/ And my "Little Book of Deep Learning" is available as a phone-formatted pdf (nearing 700k downloads!) fleuret.org/lbdl/
- Reposted by Simone Azeglio @NeurIPS[Not loaded yet]
- Yesterday's seminar was great, thanks @amlab.bsky.social for hosting! In case you missed it, you'll find this and previous seminars on our YouTube channel: www.youtube.com/@neurreps
- Reposted by Simone Azeglio @NeurIPS[Not loaded yet]
- Reposted by Simone Azeglio @NeurIPS[Not loaded yet]
- Reposted by Simone Azeglio @NeurIPS[Not loaded yet]
- Reposted by Simone Azeglio @NeurIPSWe collected lecture notes and blog posts by group members about recent topics in deep learning theory here. Hope it is useful! pehlevan.seas.harvard.edu/resources-0
- Reposted by Simone Azeglio @NeurIPS[Not loaded yet]
- Reposted by Simone Azeglio @NeurIPS[Not loaded yet]
- Reposted by Simone Azeglio @NeurIPSI'm TAing for a class and wanted to put together a (short) list of papers that are a good, accessible intro for students to get started in mech interp. An ask for the community: what papers would you add / which papers am I missing? (1/n)
- Reposted by Simone Azeglio @NeurIPS[Not loaded yet]
