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 1203.3783stat.MLcs.AIcs.LG
收藏

基于中心深度玻尔兹曼机的特征层次学习

Learning Feature Hierarchies with Centered Deep Boltzmann Machines

Korea University(高丽大学)

Grégoire Montavon, Klaus-Robert Müller

摘要:

深度玻尔兹曼机原则上是提取数据层次结构的强大模型。不幸的是,联合训练层的尝试(没有贪婪的逐层预训练)在很大程度上是不成功的。我们提出了一个修改的学习算法,最初重新中心的激活函数的输出为零。这种修改导致更好的条件海森,从而使学习更容易。我们在真实数据上测试了该算法,并证明了我们的建议,即中心深度玻尔兹曼机,学习了一个越来越抽象的表示层次和一个更好的数据生成模型。

英文摘要:

Deep Boltzmann machines are in principle powerful models for extracting the hierarchical structure of data. Unfortunately, attempts to train layers jointly (without greedy layer-wise pretraining) have been largely unsuccessful. We propose a modification of the learning algorithm that initially recenters the output of the activation functions to zero. This modification leads to a better conditioned Hessian and thus makes learning easier. We test the algorithm on real data and demonstrate that our suggestion, the centered deep Boltzmann machine, learns a hierarchy of increasingly abstract representations and a better generative model of data.

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。

↑