Home Research Papers Datasets State of the Art Pricing

Discover, visualize, and connect AI research papers. Explore the latest trends and insights in artificial intelligence research.

Product

Home
Research Papers
About

Support

Contact
Terms of Service
Privacy Policy

© 2026 Papersgraph. All rights reserved.

medical-1

Classifier calibration

3260 papers • 126 benchmarks • 313 datasets

Confidence calibration – the problem of predicting probability estimates representative of the true correctness likelihood – is important for classification models in many applications. The two common calibration metrics are Expected Calibration Error (ECE) and Maximum Calibration Error (MCE).

(Image credit: Papersgraph)

Benchmarks

These leaderboards are used to track progress in classifier-calibration-3

Trend

Dataset

Best Model

Actions

CIFAR-100

CIFAR-100

Libraries

i

Use these libraries to find classifier-calibration-3 models and implementations

Datasets

CIFAR-100

Subtasks

No subtasks available.

Most implemented papers

Masksembles for Uncertainty Estimation

P. Fua, P. Baqué, Timur M. Bagautdinov, N. Durasov•Mon Dec 14 2020

Instead of randomly dropping parts of the network as in MC-dropout, Masksemble relies on a fixed number of binary masks, which are parameterized in a way that allows to change correlations between individual models.

102

Content

Introduction Benchmarks Datasets Subtasks Libraries Papers

0

Multi-class probabilistic classification using inductive and cross Venn-Abers predictors

Valery Manokhin•Tue May 30 2017

This work presents a new approach to multi-class probability estimation by turning IVAPs and CVAPs into multiclass probabilistic predictors, which are experimentally more accurate than both uncalibrated predictors and existing calibration methods.

17 0

How Well Do Self-Supervised Models Transfer?

Timothy M. Hospedales, Henry Gouk, Linus Ericsson•Wed Nov 25 2020

It is shown that on most tasks the best self-supervised models outperform supervision, confirming the recently observed trend in the literature and finding ImageNet Top-1 accuracy to be highly correlated with transfer to many-shot recognition, but increasingly less so for few-shot, object detection and dense prediction.

307 0

Multivariate Confidence Calibration for Object Detection

Fabian Küppers, Jan Kronenberger, Amirhossein Shantia, Anselm Haselhoff•Tue Mar 31 2020

This work presents a novel framework to measure and calibrate biased (or miscalibrated) confidence estimates of object detection methods and proposes a new measure to evaluate miscalibration of object detectors.

130 0

Classifier Calibration: With Application to Threat Scores in Cybersecurity

W. Yousef, I. Traoré, William Briguglio•Mon Feb 08 2021

This article proposes expanding these classifiers’ scores to higher dimensions to boost the calibrator’s performance, and suggests multi-score calibration provides better performance than single score calibration in the majority of experiments, including the two real datasets.

5 0

Generalized and Incremental Few-Shot Learning by Explicit Learning and Calibration without Forgetting

B. Schiele, Hilde Kuehne, Anna Kukleva•Tue Aug 17 2021

This work proposes a three-stage framework that allows to explicitly and effectively address the challenges of generalized and incremental few shot learning and evaluates the proposed framework on four challenging benchmark datasets for image and video few-shot classification and obtains state-of-the-art results.

70 0

Hidden Heterogeneity: When to Choose Similarity-Based Calibration

Thomas G. Dietterich, K. Wagstaff•Wed Feb 02 2022

HH can serve as a useful diagnostic tool for identifying when local calibration methods would be beneficial, and two similarity-weighted calibration methods that can address HH by adapting locally to each test item are introduced.

1 0

No Fear of Heterogeneity: Classifier Calibration for Federated Learning with Non-IID Data

Yifan Zhang, Jiashi Feng, Romy Mi Luo, Fei Chen, Dapeng Hu, Jian Liang•Tue Jun 08 2021

A novel and simple algorithm called Classifier Calibration with Virtual Representations (CCVR), which adjusts the classifier using virtual representations sampled from an approximated gaussian mixture model, which achieves state-of-the-art performance on popular federated learning benchmarks including CIFAR-10, CIFar-100, and CINIC-10.

439 0

Class-wise and reduced calibration methods

Michael Panchenko, Anes Benmerzoug, Miguel de Benito Delgado•Thu Oct 06 2022

This work proves for several notions of calibration that solving the reduced problem minimizes the corresponding notion of miscalibration in the full problem, allowing the use of non-parametric recalibration methods that fail in higher dimensions.

3 0

Adding a benchmark result helps the community track progress.

Classifier calibration | State-of-the-Art