Explainability as statistical inference
Fuente:
arXiv
Salvato in:
| Autori principali: | , , , |
|---|---|
| Natura: | Preprint |
| Pubblicazione: |
2022
|
| Soggetti: | |
| Accesso online: | |
| Tags: |
Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
|
| _version_ | 1866914623958024192 |
|---|---|
| author | Senetaire, Hugo Henri Joseph Garreau, Damien Frellsen, Jes Mattei, Pierre-Alexandre |
| author_facet | Senetaire, Hugo Henri Joseph Garreau, Damien Frellsen, Jes Mattei, Pierre-Alexandre |
| contents | A wide variety of model explanation approaches have been proposed in recent years, all guided by very different rationales and heuristics. In this paper, we take a new route and cast interpretability as a statistical inference problem. We propose a general deep probabilistic model designed to produce interpretable predictions. The model parameters can be learned via maximum likelihood, and the method can be adapted to any predictor network architecture and any type of prediction problem. Our method is a case of amortized interpretability models, where a neural network is used as a selector to allow for fast interpretation at inference time. Several popular interpretability methods are shown to be particular cases of regularised maximum likelihood for our general model. We propose new datasets with ground truth selection which allow for the evaluation of the features importance map. Using these datasets, we show experimentally that using multiple imputation provides more reasonable interpretations. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2212_03131 |
| institution | arXiv |
| publishDate | 2022 |
| record_format | arxiv |
| spellingShingle | Explainability as statistical inference Senetaire, Hugo Henri Joseph Garreau, Damien Frellsen, Jes Mattei, Pierre-Alexandre Machine Learning Artificial Intelligence Methodology A wide variety of model explanation approaches have been proposed in recent years, all guided by very different rationales and heuristics. In this paper, we take a new route and cast interpretability as a statistical inference problem. We propose a general deep probabilistic model designed to produce interpretable predictions. The model parameters can be learned via maximum likelihood, and the method can be adapted to any predictor network architecture and any type of prediction problem. Our method is a case of amortized interpretability models, where a neural network is used as a selector to allow for fast interpretation at inference time. Several popular interpretability methods are shown to be particular cases of regularised maximum likelihood for our general model. We propose new datasets with ground truth selection which allow for the evaluation of the features importance map. Using these datasets, we show experimentally that using multiple imputation provides more reasonable interpretations. |
| title | Explainability as statistical inference |
| topic | Machine Learning Artificial Intelligence Methodology |
| url | https://arxiv.org/abs/2212.03131 |