Research map / Learning theory and algorithms / Kernels and regression theory
Learning Latent Space Models with Angular Constraints
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 |
|---|---|---|---|
| add angular constraint (ablation) | sparse coding · UIUC-Sport | accuracy (%): 87.4 -> 90.9 | Helped |
| add angular constraint (ablation) | sparse coding · Scene-15 | accuracy (%): 83.6 -> 86.5 | Helped |
| add angular constraint (ablation) | sparse coding · Caltech-256 | accuracy (%): 42.3 -> 46.1 | Helped |
| replace CS regularizer with AC | sparse coding · Caltech-256 | accuracy (%): 45.4 -> 46.1 | Helped |
| add angular constraint (ablation) | Kaldi FNN · TIMIT core test | phone error rate (%): 18.53 -> 18.41 | Helped |
| add AC instead of mutual-angle regularizer | Kaldi FNN · TIMIT core test | phone error rate (%): 18.51 -> 18.41 | Helped |
| add angular constraint (ablation) | WideResNet-28-10 · CIFAR-10 test | classification error (%): 3.89 -> 3.63 | Helped |
| add MA regularizer instead of AC | WideResNet-28-10 · CIFAR-10 test | classification error (%): 3.89 -> 3.68 | Helped |
| add angular constraint (ablation) | BIDAF · CNN QA test | accuracy (%): 76.94 -> 77.23 | Helped |
| add angular constraint (ablation) | BIDAF · DailyMail QA test | accuracy (%): 79.63 -> 79.88 | Helped |
| vary nearly-orthogonality parameter tau toward 0 (ablation) | AC-regularized FNN · TIMIT core test | phone error rate | Hurt |
| moderate tau of 0.75 versus smaller or larger (ablation) | AC-regularized FNN · TIMIT core test | phone error rate | Helped |
Closest papers by meaning
- Metric Space Magnitude for Evaluating the Diversity of Latent Representations (NeurIPS 2024 · similarity 0.85)
- Stabilize the Latent Space for Image Autoregressive Modeling: A Unified Perspective (NeurIPS 2024 · similarity 0.84)
- Compress to Impress: Efficient LLM Adaptation Using a Single Gradient Step on 100 Samples (NeurIPS 2025 · similarity 0.84)
- Linear combinations of latents in generative models: subspaces and beyond (ICLR 2025 · similarity 0.84)
- MITIGATING OVER-EXPLORATION IN LATENT SPACE OPTIMIZATION USING LES (ICML 2025 · similarity 0.83)
- GLAD: Improving Latent Graph Generative Modeling with Simple Quantization (AAAI 2025 · similarity 0.83)
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.