Beyond temperature scaling: Obtaining well-calibrated multi-class probabilities with Dirichlet calibration

Meelis Kull, Miquel Perello Nieto, Markus Kängsepp, Telmo Silva Filho, Hao Song, Peter Flach

Nov-17-2025, 17:46:41 GMT–Neural Information Processing Systems

It is easily implemented with neural nets since it is equivalent to log-transforming the uncalibrated probabilities, followed by one linear layer and softmax.

artificial intelligence, calibration, machine learning, (19 more...)

Neural Information Processing Systems

Nov-17-2025, 17:46:41 GMT

Conferences PDF

Country:
- Europe
  - Estonia > Tartu County
    - Tartu (0.04)
  - Sweden > Stockholm
    - Stockholm (0.04)
- North America > Canada (0.04)
- Oceania > Australia
  - New South Wales > Sydney (0.04)

Genre:
- Research Report
  - Experimental Study (0.69)
  - New Finding (0.94)

Technology:
- Information Technology > Artificial Intelligence > Machine Learning
  - Learning Graphical Models > Directed Networks
    - Bayesian Learning (0.68)
  - Neural Networks > Deep Learning (0.47)
  - Statistical Learning (1.00)

Duplicate Docs Excel Report

Title
Beyond temperature scaling: Obtaining well-calibrated multi-class probabilities with Dirichlet calibration

Similar Docs Excel Report more

Title	Similarity	Source
None found