AI Paper Atlas

Research map / Optimization / Implicit bias and generalization

Deep-ICE: The first globally optimal algorithm for empirical risk minimization of two-layer maxout and ReLU networks

Xi He, Yi Miao, Max A Little

ICLR 2026 · Optimization › Implicit bias and generalization · approach: Shallow ReLU network convergence

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
exact 0-1 loss optimization (Deep-ICE) instead of gradient descentrank-2 maxout · VP dataset N=704training 0-1 loss: 25 -> 16Improved
Deep-ICE K=2 with coresets instead of SVMtwo-layer maxout · UCI Caesariantest accuracy (%): 57.33 -> 88.00Improved
Deep-ICE K=2 with coresets instead of SVMtwo-layer maxout · UCI DBtest accuracy (%): 67.62 -> 81.37Improved
Deep-ICE K=2 instead of MLP K=2 trained by Adamtwo-layer maxout · UCI SStest accuracy (%): 79.65 -> 86.72Improved
Deep-ICE K=2 instead of EXACT K=2two-layer maxout · UCI Voicepathtest accuracy (%): 92.47 -> 97.59Improved
Deep-ICE K=2 instead of EXACT K=2two-layer maxout · UCI Ai4itest accuracy (%): 96.60 -> 97.82Improved
increase hidden units from K=1 to K=2 (ablation)Deep-ICE · UCI Caesariantest accuracy (%): 82.67 -> 88.00Improved
increase hidden units from K=2 to K=3 (ablation)Deep-ICE · UCI Caesariantest accuracy (%): 88.00 -> 86.67Hurt
increase hidden units from K=2 to K=3 (ablation)Deep-ICE · UCI SOtest accuracy (%): 75.33 -> 74.95No clear effect
Deep-ICE exact solutions generalize despite higher training accuracyDeep-ICE · UCI datasetstest accuracyImproved

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