How to Evaluate Behavioral Models
Fuente:
arXiv
Salvato in:
| Autori principali: | , , , |
|---|---|
| Natura: | Preprint |
| Pubblicazione: |
2023
|
| Soggetti: | |
| Accesso online: | |
| Tags: |
Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
|
| _version_ | 1866910341167841280 |
|---|---|
| author | d'Eon, Greg Greenwood, Sophie Leyton-Brown, Kevin Wright, James R. |
| author_facet | d'Eon, Greg Greenwood, Sophie Leyton-Brown, Kevin Wright, James R. |
| contents | Researchers building behavioral models, such as behavioral game theorists, use experimental data to evaluate predictive models of human behavior. However, there is little agreement about which loss function should be used in evaluations, with error rate, negative log-likelihood, cross-entropy, Brier score, and squared L2 error all being common choices. We attempt to offer a principled answer to the question of which loss functions should be used for this task, formalizing axioms that we argue loss functions should satisfy. We construct a family of loss functions, which we dub "diagonal bounded Bregman divergences", that satisfy all of these axioms. These rule out many loss functions used in practice, but notably include squared L2 error; we thus recommend its use for evaluating behavioral models. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2306_04778 |
| institution | arXiv |
| publishDate | 2023 |
| record_format | arxiv |
| spellingShingle | How to Evaluate Behavioral Models d'Eon, Greg Greenwood, Sophie Leyton-Brown, Kevin Wright, James R. Machine Learning Computer Science and Game Theory Researchers building behavioral models, such as behavioral game theorists, use experimental data to evaluate predictive models of human behavior. However, there is little agreement about which loss function should be used in evaluations, with error rate, negative log-likelihood, cross-entropy, Brier score, and squared L2 error all being common choices. We attempt to offer a principled answer to the question of which loss functions should be used for this task, formalizing axioms that we argue loss functions should satisfy. We construct a family of loss functions, which we dub "diagonal bounded Bregman divergences", that satisfy all of these axioms. These rule out many loss functions used in practice, but notably include squared L2 error; we thus recommend its use for evaluating behavioral models. |
| title | How to Evaluate Behavioral Models |
| topic | Machine Learning Computer Science and Game Theory |
| url | https://arxiv.org/abs/2306.04778 |