Who's Who of AI

jvgemert.bsky.social

85 trust researcher @jvgemert.bsky.social · 1,299 followers
AI research Vision & synthetic media Compute & infrastructure
Why they matter

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

AI signals
2
past 30d
Sources
1
distinct domains
Discusiones
7
past 30d
Latest signal
9d ago
View every signal from jvgemert.bsky.social →
Head of the Computer Vision lab; TU Delft. - Fundamental, empirical, understanding-based, Deep Learning research - Visual inductive priors for data efficiency Web: https://jvgemert.github.io/
What they're sharing

Articles & links

In 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/2610.05902

On Hyperparameter Tuning on the Test Set arxiv.org
AI Weekly's analysis →
  • A new arxiv paper by Fregonara, Viering and van Gemert tests test-set hyperparameter tuning on MNIST-1D, CIFAR-10 and three GLUE tasks.
  • The authors find the inflation effect is real and significant but frequently small relative to other sources of noise.
  • Model rankings remain essentially preserved after test-set tuning, so consistent test tuning may not invalidate benchmarks.
Read full analysis →
View on Bluesky · ♥ 23 ↻ 4 ↩ 2 · 2 from the directory shared this · 17d ago

I really enjoy finding gaps in our understanding of text-book deep learning. The RNN Homunculus: Long-range input vanishes RNN gradients. Gates (GRU, LSTM) help, but gates are themselves an RNN. Ie: gates are not sufficient. We need short-range data to learn gates first. arxiv…

The RNN Homunculus: Learning to Gate Long-Term Memory arxiv.org
AI Weekly's analysis →
  • Gated RNNs can match or exceed Mamba and xLSTM test accuracy when their gates are learned properly, the authors claim.
  • The analysis finds gated architectures suffer the vanishing gradient problem when all training samples require long-range memory.
  • Adding short-dependency samples lets gated models learn solutions that capture long dependencies and generalize across class labels.
Read full analysis →
View on Bluesky · ♥ 18 ↻ 3 ↩ 0 · 2 from the directory shared this · 9d ago
Their own posts

Recent commentary

Honered to get a selfie with our Netherlands Conference on Computer Vision (NCCV) keynote speaker! The brilliant @vickykalogeiton.bsky.social Two zealots of the Data efficiency church in the Religion of Scaling AI 🙏

View on Bluesky · ♥ 11 ↻ 0 ↩ 0 · 127d ago
Their network

In jvgemert.bsky.social's orbit

Center = jvgemert.bsky.social. 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 jvgemert.bsky.social? 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/jvgemert-bsky-social)