Adaptive joint distribution learning
Fuente:
arXiv
Salvato in:
| Autori principali: | , , |
|---|---|
| Natura: | Preprint |
| Pubblicazione: |
2021
|
| Soggetti: | |
| Accesso online: | |
| Tags: |
Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
|
| _version_ | 1866909323373838336 |
|---|---|
| author | Filipovic, Damir Multerer, Michael Schneider, Paul |
| author_facet | Filipovic, Damir Multerer, Michael Schneider, Paul |
| contents | We develop a new framework for estimating joint probability distributions using tensor product reproducing kernel Hilbert spaces (RKHS). Our framework accommodates a low-dimensional, normalized and positive model of a Radon--Nikodym derivative, which we estimate from sample sizes of up to several millions, alleviating the inherent limitations of RKHS modeling. Well-defined normalized and positive conditional distributions are natural by-products to our approach. Our proposal is fast to compute and accommodates learning problems ranging from prediction to classification. Our theoretical findings are supplemented by favorable numerical results. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2110_04829 |
| institution | arXiv |
| publishDate | 2021 |
| record_format | arxiv |
| spellingShingle | Adaptive joint distribution learning Filipovic, Damir Multerer, Michael Schneider, Paul Machine Learning Numerical Analysis 65D05, 65D15, 62G07 We develop a new framework for estimating joint probability distributions using tensor product reproducing kernel Hilbert spaces (RKHS). Our framework accommodates a low-dimensional, normalized and positive model of a Radon--Nikodym derivative, which we estimate from sample sizes of up to several millions, alleviating the inherent limitations of RKHS modeling. Well-defined normalized and positive conditional distributions are natural by-products to our approach. Our proposal is fast to compute and accommodates learning problems ranging from prediction to classification. Our theoretical findings are supplemented by favorable numerical results. |
| title | Adaptive joint distribution learning |
| topic | Machine Learning Numerical Analysis 65D05, 65D15, 62G07 |
| url | https://arxiv.org/abs/2110.04829 |