@inproceedings{ecd4c180610848f0a8a0c0bfefd7201e,
title = "NP-PROV: Neural Processes with Position-Relevant-Only Variances",
abstract = "Neural Processes (NPs) families encode distributions over functions to a latent representation given a set of context data, and decode posterior mean and variance at unknown locations. Since mean and variance are derived from the same latent space, they may fail on out-of-domain tasks where fluctuations in function values amplify the model uncertainty. We present a new member named Neural Processes with Position-Relevant-Only Variances (NP-PROV). NP-PROV hypothesizes that a target point close to a context point has small uncertainty, regardless of the function value at that position. The resulting approach derives mean and variance from a function-value-related space and a position-related-only latent space separately. Our evaluation on synthetic and real-world datasets reveals that NP-PROV can achieve state-of-the-art likelihood while retaining a bounded variance when drifts exist in the function value.",
keywords = "Neural processes, Position-relevant-only variance, Uncertainty evaluation",
author = "Xuesong Wang and Lina Yao and Xianzhi Wang and Feiping Nie and Boualem Benatallah",
note = "Publisher Copyright: {\textcopyright} 2021, Springer Nature Switzerland AG.; 22nd International Conference on Web Information Systems Engineering, WISE 2021 ; Conference date: 26-10-2021 Through 29-10-2021",
year = "2021",
doi = "10.1007/978-3-030-90888-1_11",
language = "英语",
isbn = "9783030908874",
series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",
publisher = "Springer Science and Business Media Deutschland GmbH",
pages = "129--142",
editor = "Wenjie Zhang and Lei Zou and Zakaria Maamar and Lu Chen",
booktitle = "Web Information Systems Engineering - WISE 2021 - 22nd International Conference on Web Information Systems Engineering, WISE 2021, Proceedings",
}