TE
Thomas Elsken
cs.LGstat.MLcs.AIcs.CVcs.NE
On Valency
published · living versionsW_7r4m86sk·v1 · currentpublished
Don't Skip the Baseline: A Controlled Comparison of Dynamic Expert Skipping in MoEs
with Jakob Steimle, Daniel Ehrhardt, Lukas Rinder, Daniel Cremers +1
1 version
Preprints & journals
14 papers in the corpus · 1997–2026Don't Skip the Baseline: A Controlled Comparison of Dynamic Expert Skipping in MoEs2610.28369v1 · Jakob Steimle, Thomas Elsken, Daniel Ehrhardt et al.2026 · 0 citationsarXivon Valency
Neural Architecture Search: Insights from 1000 Papers2301.08727v2 · Colin White, Mahmoud Safari, Rhea Sukthanker et al.2023 · 72 citationsarXiv
Neural Architecture Search for Dense Prediction Tasks in Computer Vision2202.07242v1 · Thomas Elsken, Arber Zela, Jan Hendrik Metzen et al.2022 · 20 citationsarXiv
Bag of Tricks for Neural Architecture Search2107.03719v1 · Thomas Elsken, Benedikt Staffler, Arber Zela et al.2021 · 4 citationsarXiv
Bag of Baselines for Multi-objective Joint Neural Architecture Search and Hyperparameter Optimization2105.01015v1 · Julia Guerrero-Viu, Sven Hauns, Sergio Izquierdo et al.2021 · 15 citationsarXiv
Neural Ensemble Search for Uncertainty Estimation and Dataset Shift2006.08573v3 · Sheheryar Zaidi, Arber Zela, Thomas Elsken et al.2020 · 42 citationsarXiv
Meta-Learning of Neural Architectures for Few-Shot Learning1911.11090v3 · Thomas Elsken, Benedikt Staffler, Jan Hendrik Metzen et al.2019 · 171 citations2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)
Automated design of error-resilient and hardware-efficient deep neural networks1909.13844v1 · Christoph Schorn, Thomas Elsken, Sebastian Vogel et al.2019 · 39 citationsarXiv
Understanding and Robustifying Differentiable Architecture Search1909.09656v2 · Arber Zela, Thomas Elsken, Tonmoy Saikia et al.2019 · 279 citationsarXiv
Neural Architecture Search: A Survey1808.05377v3 · Thomas Elsken, Jan Hendrik Metzen, Frank Hutter2018 · 1,762 citationsJournal of Machine Learning Research 20 (2019) 1-21
Efficient Multi-objective Neural Architecture Search via Lamarckian Evolution1804.09081v4 · Thomas Elsken, Jan Hendrik Metzen, Frank Hutter2018 · 300 citationsarXiv
Simple And Efficient Architecture Search for Convolutional Neural Networks1711.04528v1 · Thomas Elsken, Jan-Hendrik Metzen, Frank Hutter2017 · 207 citationsarXiv
Smaller nets may perform better: special transfer functions.12662674 · Elsken, T2019 · 2 citationsNeural networks : the official journal of the International Neural Network Society. 1999;12(4-5):627-645
Even on Finite Test Sets Smaller Nets may Perform Better.12662534 · ELSKEN, THOMAS2019 · 7 citationsNeural networks : the official journal of the International Neural Network Society. 1997;10(2):369-385
Career total: 29 works. 14 are in this corpus.
Profile built from the corpus for this byline.
Author records are still filling in while the Hub is in alpha. If this is your page, you'll be able to claim it soon. Spot a mistake? Tell us.