"Learning (Approximately) Equivariant Networks via Constrained Optimization" will be an Oral ๐ฃ๏ธ at NeurIPS! (w/ @Mniepert & @lfochamon) ACE goes beyond fixed equivariance, augmentations, and regularizers by learning from data when to enforce symmetry and when to break it. ๐งต2/9
@PMinervini I'm wondering how much of this is due to the language barrier for international employees/founders v.s. other more structural reasons like proximity to good universities or a different VC culture from the rest of Europe.
โฐ The CoLLAs abstract deadline is only 10 days away!
We invite researchers to explore all facets of ML adaptation, from incorporating new capabilities during continuous training to efficiently removing outdated or harmful data.
- ๐๐ฏ๐๐๐ฟ๐ฎ๐ฐ๐ ๐๐ฒ๐ฎ๐ฑ๐น๐ถ๐ป๐ฒ: April 10, 2026
- ๐ฆ๐๐ฏ๐บ๐ถ๐๐๐ถ๐ผ๐ป ๐๐ฒ๐ฎ๐ฑ๐น๐ถ๐ป๐ฒ: April 15, 2026
- ๐๐ผ๐ป๐ณ๐ฒ๐ฟ๐ฒ๐ป๐ฐ๐ฒ ๐๐ฎ๐๐ฒ๐: Sep 14โ17, 2026
๐ Accepted papers will be published in the Proceedings of Machine Learning Research (PMLR).
๐ ๐๐ผ๐ฟ ๐ณ๐๐น๐น ๐ฑ๐ฒ๐๐ฎ๐ถ๐น๐ ๐ผ๐ป ๐๐ต๐ฒ ๐๐ฎ๐น๐น ๐ณ๐ผ๐ฟ ๐ฃ๐ฎ๐ฝ๐ฒ๐ฟ๐: https://t.co/tC3mhTzZqf
Weโre bringing the CoLLAs 2026 community to Bucharest, Romania ๐ท๐ด next September!
๐ง Focus: Adaptation, Continual Learning, Unlearning
๐ Submit by mid-April
Chat about CoLLAs topics or must-see spots in Bucharest at #NeurIPS2025 with Sarath, Razvan, and Eleni! ๐
I respect that @iclr_conf had to respond to the OR leak, but I disagree with resetting scores. Many students worked hard on rebuttals and improved their papers in good faith. I hope the organizers reconsider and revert the reset. If you agree, feel free to retweet.
@kchonyc@orf_bnw Also happened to me yesterday with Nano Banana. ChatGPT managed to generate one with a transparent background, but the Nano Banana one was nicer.
@Mniepert@lfochamon ChronoGraph (w/ A. Lutu, I. Pintilie & @ilarele) is a real-world dataset of microservice telemetry that links multivariate time series with a dependency graph and incident labels, enabling evaluation of forecasting and anomaly detection methods under propagation effects.
๐งต8/9
@Mniepert@lfochamon Inspired by JEPA, C-FREE (w/ B. Ariguib & @mniepert) uses a predictive objective over egonets spanning 2D and 3D graphs, avoiding negatives, reconstructions, and augmentations. This simple multimodal design delivers strong results from low- to large-scale regimes. ๐ฌ ๐งต6/9
"Learning (Approximately) Equivariant Networks via Constrained Optimization" will be an Oral ๐ฃ๏ธ at NeurIPS! (w/ @Mniepert & @lfochamon) ACE goes beyond fixed equivariance, augmentations, and regularizers by learning from data when to enforce symmetry and when to break it. ๐งต2/9
Looking forward to #NeurIPS next week! Iโm presenting three works across equivariant learning, molecular graph pretraining, and real-world graph-based time series:
โข ACE (NeurIPS Oral)
โข C-FREE (NPGML Poster)
โข ChronoGraph (BERT2S Oral)
Details in the thread โฌ๏ธ๐งต1/9
@Mniepert@lfochamon ChronoGraph (w/ A. Lutu, I. Pintilie & @ilarele) is a real-world dataset of microservice telemetry that links multivariate time series with a dependency graph and incident labels, enabling evaluation of forecasting and anomaly detection methods under propagation effects.
๐งต8/9
@PetarV_93 Btw, did you know that the studio hired Lorien Testard (main composer) because they just stumbled upon his work on some random forum? :) Best outcome for everyone involved.