arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 1301.3568stat.MLcs.LG
收藏

联合训练深度玻尔兹曼机分类

Joint Training Deep Boltzmann Machines for Classification

Université de Montréal(蒙特利尔大学)

Ian J. Goodfellow, Aaron Courville, Yoshua Bengio

摘要:

我们介绍了一种新的方法来训练深度玻尔兹曼机联合。训练DBM的现有方法需要初始学习通道,该初始学习通道一次一层地训练模型,或者在分类任务上表现不佳。在我们的方法中,我们同时训练DBM的所有层,使用一种新的训练过程称为多预测训练。由此产生的模型可以被解释为一个单一的生成模型训练,以最大限度地提高变分近似的广义pseudolikket,或作为一个家庭的经常性网络,共享参数,并可以近似平均在一起使用一种新的技术,我们称之为多推理技巧。我们表明,我们的方法进行竞争力的分类,并优于以前的方法在精度的近似推理和分类与丢失的输入。

英文摘要:

We introduce a new method for training deep Boltzmann machines jointly. Prior methods of training DBMs require an initial learning pass that trains the model greedily, one layer at a time, or do not perform well on classification tasks. In our approach, we train all layers of the DBM simultaneously, using a novel training procedure called multi-prediction training. The resulting model can either be interpreted as a single generative model trained to maximize a variational approximation to the generalized pseudolikelihood, or as a family of recurrent networks that share parameters and may be approximately averaged together using a novel technique we call the multi-inference trick. We show that our approach performs competitively for classification and outperforms previous methods in terms of accuracy of approximate inference and classification with missing inputs.

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。

↑