Research map / Learning theory and algorithms / Kernels and regression theory
Inductive Two-Layer Modeling with Parametric Bregman Transfer
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 changed | Where | Measured | Effect |
|---|---|---|---|
| inductive CVX-IN instead of transductive CVX-TR | CVX-IN · Letter, G241N, XOR, CIFAR-SM (100/100) | mean test error: G241N 24.2 vs 23.1; Letter 4.8 vs 4.9 | Mixed |
| CVX-IN instead of CVX-TR | CVX-IN · G241N (1000/1000) | mean test error: 13.0 vs 15.1 | Helped |
| convex two-layer CVX-IN instead of FFNN | CVX-IN · G241N (100/100) | mean test error: 31.9 -> 24.2 | Helped |
| CVX-IN instead of FFNN | CVX-IN · CIFAR-SM (100/100) | mean test error: 31.0 -> 21.2 | Helped |
| CVX-IN instead of FFNN | CVX-IN · G241N (1000/1000) | mean test error: 24.5 -> 13.0 | Helped |
| CVX-IN instead of FFNN | CVX-IN · XOR | mean test error: 0 vs 0 | No clear effect |
| CVX-IN instead of LOCAL | CVX-IN · XOR (100/100) | mean test error: 27.0 -> 0 | Helped |
| CVX-IN instead of Ker-CVX | CVX-IN · Letter (200/200) | mean test error: 4.5 vs 5.1 | Hurt |
| CVX-IN instead of Ker-CVX | CVX-IN · CIFAR-SM (100/100) | mean test error: 27.7 -> 21.2 | Helped |
| increase training set 100 to 200 | all methods · CIFAR-SM | mean test error: 21.2 -> 22.6 (CVX-IN) | Hurt |
| increase training size 200 to 2000 | CVX-IN · Letter | mean test error: 5.1 -> 1.0 | Helped |
| min-min alternating local optimization | CVX-IN · CIFAR-10 2000 examples | training time: 2.5 hours | Helped |
Closest papers by meaning
- Function Encoders: A Principled Approach to Transfer Learning in Hilbert Spaces (ICML 2025 · similarity 0.85)
- Simplicity Bias and Optimization Threshold in Two-Layer ReLU Networks (ICML 2025 · similarity 0.84)
- Local Loss Optimization in the Infinite Width: Stable Parameterization of Predictive Coding Networks and Target Propagation (ICLR 2025 · similarity 0.84)
- Convex Formulations for Training Two-Layer ReLU Neural Networks (ICLR 2025 · similarity 0.84)
- On The Concurrence of Layer-wise Preconditioning Methods and Provable Feature Learning (ICML 2025 · similarity 0.84)
- Convex Relaxations of ReLU Neural Networks Approximate Global Optima in Polynomial Time (ICML 2024 · similarity 0.84)
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.