arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2501.08285cs.LGcs.CV
收藏

贝叶斯神经网络能否显式建模输入不确定性?

Can Bayesian Neural Networks Explicitly Model Input Uncertainty?

University of Groningen(格罗宁根大学)

Matias Valdenegro-Toro, Marco Zullich

AI总结:

针对机器学习输入不确定性常被忽略的问题,构建双输入贝叶斯神经网络,测试多种近似方法后发现仅Ensembles、Flipout等部分方法可有效建模输入不确定性。

摘要:

机器学习模型的输入可能伴随噪声或不确定性,但这类因素常被忽略、未被建模。目前尚不清楚贝叶斯神经网络及其近似方法能否考量输入中的不确定性。本文构建了一个双输入(均值与标准差)贝叶斯神经网络,在Ensembles(集成)、MC-Dropout(蒙特卡洛随机失活)、Flipout等不同方法下评估其输入不确定性估计能力。结果表明,近似贝叶斯神经网络的不确定性估计方法中仅部分可建模输入不确定性,尤以Ensembles和Flipout表现突出。

英文摘要:

Inputs to machine learning models can have associated noise or uncertainties, but they are often ignored and not modelled. It is unknown if Bayesian Neural Networks and their approximations are able to consider uncertainty in their inputs. In this paper we build a two input Bayesian Neural Network (mean and standard deviation) and evaluate its capabilities for input uncertainty estimation across different methods like Ensembles, MC-Dropout, and Flipout. Our results indicate that only some uncertainty estimation methods for approximate Bayesian NNs can model input uncertainty, in particular Ensembles and Flipout.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑