Research map / Learning theory and algorithms / Prediction and decision losses
Loss Decomposition for Fast Learning in Large Output Spaces
Ian En-Hsu Yen, Satyen Kale, Felix Yu, Daniel Holtmann-Rice, Sanjiv Kumar, Pradeep Ravikumar
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 |
|---|---|---|---|
| decomposed search (Decomp-MIPS) instead of plain MIPS for gradient approximation | max-margin loss training of last layer · MegaFace multiclass | test accuracy over training time | Improved |
| decomposed search (Decomp-MIPS) instead of plain MIPS | max-margin loss on WikiLSHTC · WikiLSHTC multilabel | training accuracy per epoch | Improved |
| sampled softmax instead of search-based gradients | softmax loss · MegaFace multiclass | test accuracy over training time | Hurt |
| sampled softmax instead of search-based gradients | softmax loss · WikiLSHTC multilabel | training accuracy | Hurt |
| decomposed search with MIPS-based gradients on skip-gram | skip-gram word embedding · BillionW | skip-gram loss vs time | Improved |
Closest papers by meaning
- PD-Sparse : A Primal and Dual Sparse Approach to Extreme Multiclass and Multilabel Classification (ICML 2016 · similarity 0.87)
- Multiclass Boosting: Margins, Codewords, Losses, and Algorithms (JMLR 2019 · similarity 0.86)
- Multiclass Loss Geometry Matters for Generalization of Gradient Descent in Separable Classification (NeurIPS 2025 · similarity 0.85)
- Loss Functions and Operators Generated by f-Divergences (ICML 2025 · similarity 0.84)
- Trading off Consistency and Dimensionality of Convex Surrogates for Multiclass Classification (NeurIPS 2024 · similarity 0.84)
- Understanding the Training Speedup from Sampling with Approximate Losses (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.
