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 2003.00952cs.LGstat.ML
收藏

最大均值离散正则化贝叶斯神经网络

Bayesian Neural Networks With Maximum Mean Discrepancy Regularization

Sapienza University of Rome(罗马第一大学)

Jary Pomponi, Simone Scardapane, Aurelio Uncini

摘要:

贝叶斯神经网络(BNN)被训练以优化其权重的整个分布而不是单个集合,在例如以下方面具有显著的优点:可解释性、多任务学习和校准。由于所产生的优化问题的棘手性,大多数BNN要么通过Monte Carlo方法进行采样,要么通过最小化变分近似的合适证据下限(ELBO)进行训练。在本文中,我们提出了后者的一个变体,其中我们取代了Kullback-Leibler发散的ELBO项与最大平均离散(MMD)估计,最近的工作在变分推理的启发。在激励我们的建议的基础上的MMD项的属性,我们继续展示了一些经验的优势,所提出的配方在国家的最先进的。特别是,我们的BNN实现更高的精度在多个基准,包括几个图像分类任务。此外,它们对权重的先验选择更鲁棒,并且它们被更好地校准。作为第二个贡献,我们提供了一个新的公式来估计给定预测的不确定性,表明它以更强大的方式对抗对抗攻击和在其输入上注入噪声,与更经典的标准(如微分熵)相比。

英文摘要:

Bayesian Neural Networks (BNNs) are trained to optimize an entire distribution over their weights instead of a single set, having significant advantages in terms of, e.g., interpretability, multi-task learning, and calibration. Because of the intractability of the resulting optimization problem, most BNNs are either sampled through Monte Carlo methods, or trained by minimizing a suitable Evidence Lower BOund (ELBO) on a variational approximation. In this paper, we propose a variant of the latter, wherein we replace the Kullback-Leibler divergence in the ELBO term with a Maximum Mean Discrepancy (MMD) estimator, inspired by recent work in variational inference. After motivating our proposal based on the properties of the MMD term, we proceed to show a number of empirical advantages of the proposed formulation over the state-of-the-art. In particular, our BNNs achieve higher accuracy on multiple benchmarks, including several image classification tasks. In addition, they are more robust to the selection of a prior over the weights, and they are better calibrated. As a second contribution, we provide a new formulation for estimating the uncertainty on a given prediction, showing it performs in a more robust fashion against adversarial attacks and the injection of noise over their inputs, compared to more classical criteria such as the differential entropy.

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。

↑