Research map / Learning theory and algorithms / Prediction and decision losses
Large Margin Discriminant Dimensionality Reduction in Prediction Space
Mohammad Saberian, Jose Costa Pereira, Can Xu, Jian Yang, Nuno Nvasconcelos
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 |
|---|---|---|---|
| replace fixed codewords with learned codewords (LADDER vs MCBoost) (ablation) | LADDER · Traffic sign, d=2 | error rate: 0.44 ± 0.03 -> 0.21 ± 0.02 | Improved |
| learn codewords instead of predefined ones, small embedding dimension (ablation) | LADDER · Traffic sign, d in [2,27] | error rate | Improved |
| compare against PCA, LDA, ProbPCA, KernelPCA, LPP, NPE then MCBoost classifier | LADDER · Traffic sign, d in [2,27] | error rate | Improved |
| learned codewords vs MCBoost fixed codewords for hashing (ablation) | LADDER · CIFAR-10 | mAP: 0.200 -> 0.224 | Improved |
| hash length 8 bits: LADDER vs KSH | LADDER · CIFAR-10 8 bits | mAP: 0.237 -> 0.224 | Hurt |
| hash length 10 bits: LADDER vs KSH | LADDER · CIFAR-10 10 bits | mAP: 0.252 -> 0.270 | Improved |
| hash length 12 bits: LADDER vs KSH | LADDER · CIFAR-10 12 bits | mAP: 0.253 -> 0.266 | Improved |
| replace PCA with LADDER embedding before Fisher Vectors | conv5+FV pipeline · MIT Indoor | accuracy: 52.9% -> 55.2% | Improved |
| replace PCA with plain MCBoost predictor embedding | conv5+FV pipeline · MIT Indoor | accuracy: 52.9% -> 52.8% | No clear effect |
| MCBoost vs ITQ unsup. in hashing (fixed codewords) | MCBoost · CIFAR-10 10 bits | mAP: 0.159 -> 0.250 | Improved |
Closest papers by meaning
- Multiclass Boosting: Margins, Codewords, Losses, and Algorithms (JMLR 2019 · similarity 0.86)
- A Forward Approach for Sufficient Dimension Reduction in Binary Classification (JMLR 2022 · similarity 0.86)
- Learning Compact Discriminant Representation via Low-Rank Bilinear Pooling (TPAMI 2025 · similarity 0.86)
- Multi-Class Support Vector Machine with Maximizing Minimum Margin (AAAI 2024 · similarity 0.85)
- Learning Towards The Largest Margins (ICLR 2022 · similarity 0.85)
- Deep Principal Support Vector Machines for Nonlinear Sufficient Dimension Reduction (ICML 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.
