Interpreting Predictive Probabilities: Model Confidence or Human Label Variation?
Fuente:
arXiv
Enregistré dans:
| Auteurs principaux: | , , , |
|---|---|
| Format: | Preprint |
| Publié: |
2024
|
| Sujets: | |
| Accès en ligne: | |
| Tags: |
Ajouter un tag
Pas de tags, Soyez le premier à ajouter un tag!
|
| _version_ | 1866929256561377280 |
|---|---|
| author | Baan, Joris Fernández, Raquel Plank, Barbara Aziz, Wilker |
| author_facet | Baan, Joris Fernández, Raquel Plank, Barbara Aziz, Wilker |
| contents | With the rise of increasingly powerful and user-facing NLP systems, there is growing interest in assessing whether they have a good representation of uncertainty by evaluating the quality of their predictive distribution over outcomes. We identify two main perspectives that drive starkly different evaluation protocols. The first treats predictive probability as an indication of model confidence; the second as an indication of human label variation. We discuss their merits and limitations, and take the position that both are crucial for trustworthy and fair NLP systems, but that exploiting a single predictive distribution is limiting. We recommend tools and highlight exciting directions towards models with disentangled representations of uncertainty about predictions and uncertainty about human labels. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2402_16102 |
| institution | arXiv |
| publishDate | 2024 |
| record_format | arxiv |
| spellingShingle | Interpreting Predictive Probabilities: Model Confidence or Human Label Variation? Baan, Joris Fernández, Raquel Plank, Barbara Aziz, Wilker Computation and Language With the rise of increasingly powerful and user-facing NLP systems, there is growing interest in assessing whether they have a good representation of uncertainty by evaluating the quality of their predictive distribution over outcomes. We identify two main perspectives that drive starkly different evaluation protocols. The first treats predictive probability as an indication of model confidence; the second as an indication of human label variation. We discuss their merits and limitations, and take the position that both are crucial for trustworthy and fair NLP systems, but that exploiting a single predictive distribution is limiting. We recommend tools and highlight exciting directions towards models with disentangled representations of uncertainty about predictions and uncertainty about human labels. |
| title | Interpreting Predictive Probabilities: Model Confidence or Human Label Variation? |
| topic | Computation and Language |
| url | https://arxiv.org/abs/2402.16102 |