AI Paper Atlas

Research map / Learning theory and algorithms / Sample complexity

Learning Distributions Generated by Single-Layer ReLU Networks in the Presence of Arbitrary Outliers

Saikiran Bulusu, Geethu Joseph, M. Cenk Gursoy, Pramod Varshney

NeurIPS 2022 · Learning theory and algorithms › Sample complexity · approach: Learning under log-concave distributions · 0 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
Add median filter to GD (ablation)GD estimator · Simulation p=0.95, d=5, N=20000runtime: 17.73 s -> 34.44 sHurt
Use trimmed-mean filter instead of median filter in GDGD estimator · Simulationruntime: 34.44 s -> 60.78 sHurt
Use SGD instead of GD (median filter)Median-filter estimator · Simulationruntime: 34.44 s -> 2.11 sImproved
Use SGD instead of GD (trimmed mean)Trimmed-mean estimator · Simulationruntime: 60.78 s -> 3.62 sImproved
Add filter (median or trimmed mean) to GD or SGDEstimators with outliers · Simulation, varying pestimation errorImproved
Median-based instead of trimmed-mean-based filterGD · Simulationestimation errorImproved
Use SGD instead of GDFiltered estimators · Simulation, Figure 2estimation errorHurt
Increase uncorrupted probability p towards 1All non-oracle schemes · Simulationestimation errorImproved
Increase number of samples N without a filterGD/SGD without filter · Simulationestimation errorNo clear effect
Increase dimension d up to 500Proposed schemes and oracle · Simulationestimation errorHurt

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy