Who's Who of AI

Diane Larlus

90 trust @dlarlus.bsky.social · 1,495 followers
AI research Vision & synthetic media Compute & infrastructure
Why they matter

Directory member with public evidence across AI research, Vision & synthetic media, Compute & infrastructure.

AI signals
4
past 30d
Sources
4
distinct domains
Discussions
5
past 30d
Latest signal
1d ago
View every signal from Diane Larlus →
Computer Vision & Machine Learning researcher at NAVER LABS europe she/her - https://dlarlus.github.io/
What they're sharing

Articles & links

We just released our #ECCV2026 paper on Model Merging for Computer Vision 🎓 arxiv.org/abs/2604.12935 Joint work w @pdejorge.bsky.social Cesar De Souza @bjoernmichele.bsky.social @mbsariyildiz.bsky.social @weinzaepfelp.bsky.social Florent Perronnin & @skamalas.bsky.social See P…

Task Alignment: A Simple Proxy for Practical Model Merging Across Diverse Vision Tasks arxiv.org
AI Weekly's analysis →
  • A NAVER Labs Europe paper accepted at ECCV 2026 introduces a 'task alignment proxy' for selecting model-merging hyperparameters without training the decoder for each candidate.
  • The method targets heterogeneous vision tasks with trainable decoders, moving beyond the CLIP image classification setup that dominated prior model merging work.
  • The authors claim the proxy speeds up hyperparameter selection by orders of magnitude while retaining downstream performance.
Read full analysis →
View on Bluesky · ♥ 12 ↻ 5 ↩ 1 · 3 from the directory shared this · 40d ago

Our #ECCV2026 IDeaL paper tackles Data-Free Multi-Teacher Distillation 🎓 arxiv.org/abs/2608.24759 see Bülent's thread below 🔽 joint work w @fyavuz1.bsky.social @mbsariyildiz.bsky.social at @naverlabseurope.bsky.social

IDeaL: Data-Free Multi-Teacher Distillation via Improved Dead Leaves arxiv.org
AI Weekly's analysis →
  • IDeaL generates teacher-specific synthetic samples using decorrelation losses at patch and image levels, removing the need for the teachers' original training images.
  • With a 1K-image budget, students trained on IDeaL samples match or surpass those trained on a 1K-image subset of ImageNet.
  • The paper, by Feyza Yavuz, Mert Bülent Sarıyıldız and Diane Larlus, is accepted at ECCV 2026.
Read full analysis →
View on Bluesky · ♥ 17 ↻ 3 ↩ 0 · 2 from the directory shared this · 11d ago
Their own posts

Recent commentary

Panel at #ECCV2026 with Dima Damen, Yilun Du and Viorica Patraucean on ''Can AI Build New Knowledge?'' @dimadamen.bsky.social @yilundu.bsky.social

View on Bluesky · ♥ 0 ↻ 0 ↩ 2 · 18h ago

Interested in merging back multiple fine-tuned models ? Drop by our #ECCV2026 poster 11 !

View on Bluesky · ♥ 4 ↻ 0 ↩ 0 · 1d ago
Their network

In Diane Larlus's orbit

Center = Diane Larlus. Left = members they follow (green edges). Right = members who follow them (blue edges). Top = mutual follows (orange edges, slightly larger). Drag any node to reposition; click to open that profile.

Are you Diane Larlus? Show it.

Add the Who’s Who of AI badge to your site or bio. It links back to this profile.

Listed in AI Weekly's Who's Who of AI

Markdown: [![Listed in AI Weekly's Who's Who of AI](https://aiweekly.co/modules/custom/aiweekly_whoswho/images/whoswho-badge.svg)](https://aiweekly.co/whos-who/person/dlarlus-bsky-social)