Vector Scaling
Vector Scaling is a post-hoc calibration method that belongs to the class of parametric scaling techniques, distinguished by its application of a learned weight vector and bias term to the logit outputs of a model in order to optimize the predicted probability distribution for improved alignment with true class likelihoods.
13
Publications
653
Citations
35
Authors
21
Institutions
Publications per year
2017–2026
1
Parents
13
1–5 of 13
35
3
h-index
2
h-index
2
h-index
1
h-index
1
h-index
1–5 of 35
21
University of Michigan
Ann Arbor, United States
2 h-index
2 publications
71 citations
University of California San Diego
San Diego, United States
2 h-index
4 publications
622 citations
Microsoft Research Asia (China)
Beijing, China
1 h-index
1 publications
30 citations

Centre National de la Recherche Scientifique
Paris, France
1 h-index
3 publications
36 citations
Queen's University
Kingston, Canada
1 h-index
2 publications
102 citations
Duke University
Durham, United States
1 h-index
2 publications
180 citations

Pennsylvania State University
State College, United States
1 h-index
3 publications
42 citations
Arizona State University
Tempe, United States
1 h-index
1 publications
57 citations
1–8 of 21
1 publications · 51 citations
1
h-index
TVCG · 1 publications · 12 citations
1
h-index
Conference · 1 publications · 14 citations
1
h-index
1 publications · 24 citations
1
h-index
1 publications · 14 citations
1
h-index
1–5