Research map / Learning theory and algorithms / Kernels and regression theory
Convex Representation Learning for Generalized Invariance in Semi-Inner-Product Space
Yingyi Ma, Vignesh Ganapathiraman, Yaoliang Yu, Xinhua Zhang
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 changed | Where | Measured | Effect |
|---|---|---|---|
| Use sipsf representations instead of vanilla mixup | Gaussian RKHS mixup classifier · MNIST, n=500, p=4n | test accuracy: 91.40\scriptstyle\pm1.04 -> 92.11\scriptstyle\pm1.01 | Improved |
| Use sipsf representations instead of vanilla mixup | Gaussian RKHS mixup classifier · USPS, n=1000, p=4n | test accuracy: 95.32\scriptstyle\pm1.13 -> 95.67\scriptstyle\pm0.96 | Improved |
| Use sipsf representations instead of vanilla mixup | Gaussian RKHS mixup classifier · Fashion MNIST, n=500, p=4n | test accuracy: 81.72\scriptstyle\pm1.96 -> 82.52\scriptstyle\pm1.49 | Improved |
| Use sipsf representations instead of vanilla mixup | Gaussian RKHS mixup classifier · Fashion MNIST, n=1000, p=4n | test accuracy: 83.69\scriptstyle\pm1.31 -> 84.34\scriptstyle\pm1.31 | Improved |
| Use sipsf with label-logic regularization instead of the hierarchical SVM | Structured multilabel prediction · Enron, 100/100 train/test | accuracy: 93.1 -> 96.2 | Improved |
| Use sipsf with label-logic regularization instead of the hierarchical SVM | Structured multilabel prediction · Reuters, 200/200 train/test | accuracy: 97.3 -> 97.2 | Hurt |
| Use sipsf with label-logic regularization instead of the hierarchical SVM | Structured multilabel prediction · WIPO, 500/500 train/test | accuracy: 98.7 -> 98.4 | Hurt |
| Use Euclidean embedding with the max transformation-derivative penalty instead of ordinary SVM | MNIST binary classification · 4 vs 9 | test accuracy: 97.1 -> 97.8 | Improved |
| Use Euclidean embedding with the max transformation-derivative penalty instead of ordinary SVM | MNIST binary classification · 2 vs 3 | test accuracy: 98.4 -> 98.9 | Improved |
Closest papers by meaning
- On the Global Optima of Kernelized Adversarial Representation Learning (ICCV 2019 · similarity 0.88)
- Learning with Exact Invariances in Polynomial Time (ICML 2025 · similarity 0.87)
- From Bricks to Bridges: Product of Invariances to Enhance Latent Space Communication (ICLR 2024 · similarity 0.86)
- Learning Conditional Invariances through Non-Commutativity (ICLR 2024 · similarity 0.86)
- Guarantees for Nonlinear Representation Learning: Non-identical Covariates, Dependent Data, Fewer Samples (ICML 2024 · similarity 0.86)
- Generalization Guarantees for Representation Learning via Data-Dependent Gaussian Mixture Priors (ICLR 2025 · similarity 0.85)
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.
