Research map / Learning theory and algorithms / Prediction and decision losses
Loss Functions for Top-k Error: Analysis and Insights
Check 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 |
|---|---|---|---|
| Use softmax loss instead of optimizing a loss for one specific k | Multiclass classification · 11 datasets | top-k performance across k | Helped |
| Use a proposed top-k loss instead of softmax for a specific k | Multiclass classification · 11 datasets | top-k performance at the target k | Helped |
| Use smooth top-k hinge losses | Top-k hinge optimization · Scaling experiments | training convergence speed | Helped |
| Use smooth top-1 SVM instead of nonsmooth top-1 SVM | ILSVRC 2012 · ILSVRC 2012 | convergence rate | Helped |
| Use smooth top-1 SVM instead of softmax | ILSVRC 2012 · ILSVRC 2012 | training cost | Helped |
| Discard the highest-scoring predictions from the loss | Truncated top-k entropy loss · Top-k classification | loss when top-k error is zero | Helped |
Closest papers by meaning
- Stochastic smoothing of the top-K calibrated hinge loss for deep imbalanced classification (ICML 2022 · similarity 0.90)
- On the consistency of top-k surrogate losses (ICML 2020 · similarity 0.89)
- Smooth Loss Functions for Deep Top-k Classification (ICLR 2018 · similarity 0.89)
- ST$_k$: A Scalable Module for Solving Top-k Problems (NeurIPS 2024 · similarity 0.85)
- Cardinality-Aware Set Prediction and Top-$k$ Classification (NeurIPS 2024 · similarity 0.84)
- Decoupled Kullback-Leibler Divergence Loss (NeurIPS 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.