Research map / Learning theory and algorithms / Prediction and decision losses
On the Characterization of a Class of Fisher-Consistent Loss Functions and its Application to Boosting
Matey Neykov, Jun S. Liu, Tianxi Cai
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 generic Fisher-consistent loss boosting (Logistic) instead of SAMME | boosting with bag of weak learners · Simulated l1 spheres, 6 classes | test misclassification rate (%): 28.4% -> 25.1% | Improved |
| use LogLog loss instead of SAMME | boosting · Simulated l1 spheres, 6 classes | test misclassification rate (%): 28.4% -> 25.2% | Improved |
| use Logistic loss instead of SAMME | boosting · Simulated l2 spheres, 6 classes | test misclassification rate (%): 28.9% -> 25.4% | Improved |
| use LogLog loss instead of SAMME | boosting · Simulated l2 spheres, 3 classes | test misclassification rate (%): 8.2% -> 7.7% | Improved |
| use Logistic loss instead of SAMME | boosting · UCI Image Segmentation | test misclassification rate (%): 5.09% -> 5.00% | Improved |
| use Logistic loss instead of SAMME | boosting · UCI Ecoli | test misclassification rate (%): 22.45% -> 19.49% | Improved |
| use Logistic loss instead of SAMME | boosting · UCI Red Wine Quality | test misclassification rate (%): 30.14% -> 27.36% | Improved |
| use LogLog loss instead of SAMME | boosting · UCI Seeds | test misclassification rate (%): 4.55% -> 3.64% | Improved |
| use LogLog loss instead of SAMME | boosting · UCI LED Display | test misclassification rate (%): 29.36% -> 27.00% | Improved |
| use Logistic loss instead of SAMME | boosting · EMR diabetic neuropathy (3 classes) | test misclassification rate (%): 33.33% -> 31.67% | Improved |
| use boosting with Logistic loss instead of SVM | boosting · Simulated l1 spheres, 6 classes | test misclassification rate (%): 46.6% -> 25.1% | Improved |
| use boosting with Logistic loss instead of LASSO logistic regression | boosting · EMR diabetic neuropathy (3 classes) | test misclassification rate (%): 37.00% -> 31.67% | Improved |
Closest papers by meaning
- On the Consistency of Ordinal Regression Methods (JMLR 2017 · similarity 0.88)
- von Mises-Fisher Loss: An Exploration of Embedding Geometries for Supervised Learning (ICCV 2021 · similarity 0.87)
- Realizable $H$-Consistent and Bayes-Consistent Loss Functions for Learning to Defer (NeurIPS 2024 · similarity 0.86)
- Trading off Consistency and Dimensionality of Convex Surrogates for Multiclass Classification (NeurIPS 2024 · similarity 0.86)
- A Universal Growth Rate for Learning with Smooth Surrogate Losses (NeurIPS 2024 · similarity 0.85)
- Robust Loss Functions for Training Decision Trees with Noisy Labels (AAAI 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.
