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 1606.02556cs.CVcs.AI
收藏

DISCO网:相异系数网络

DISCO Nets: DISsimilarity COefficient Networks

University of Oxford(牛津大学);Microsoft Research Cambridge(微软研究院剑桥分院)

Diane Bouchacourt, M. Pawan Kumar, Sebastian Nowozin

摘要:

我们提出了一种新的概率模型,我们称之为相异系数网络(DISCO网)。DISCO网络允许我们从由神经网络参数化的后验分布中有效地采样。在训练过程中,DISCO网络通过最小化真实分布和估计分布之间的相异系数来学习。这使我们能够根据与手头任务相关的损失量身定制训练。我们的经验表明,(i)通过对输出值的不确定性建模,DISCO Nets优于等效的非概率预测网络,(ii)DISCO Nets准确地对输出的不确定性建模,优于基于深度神经网络的现有概率模型。

英文摘要:

We present a new type of probabilistic model which we call DISsimilarity COefficient Networks (DISCO Nets). DISCO Nets allow us to efficiently sample from a posterior distribution parametrised by a neural network. During training, DISCO Nets are learned by minimising the dissimilarity coefficient between the true distribution and the estimated distribution. This allows us to tailor the training to the loss related to the task at hand. We empirically show that (i) by modeling uncertainty on the output value, DISCO Nets outperform equivalent non-probabilistic predictive networks and (ii) DISCO Nets accurately model the uncertainty of the output, outperforming existing probabilistic models based on deep neural networks.

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。

↑