TVAE: Triplet-Based Variational Autoencoder using Metric Learning
TVAE: Triplet-Based Variational Autoencoder using Metric Learning
复制标题
TVAE：使用度量学习的基于三元组的变分自动编码器
DOI：
--
复制
发表时间：
2018
期刊：
影响因子：
--
通讯作者：
D. Rubin
中科院分区：
文献类型：
--
作者：
Haque Ishfaq;A. Hoogi;D. Rubin
深度度量学习已被证明在学习语义表示和编码信息方面非常有效，这些信息可用于测量数据相似性，这依赖于从度量学习中学习到的嵌入。同时，变分自编码器（VAE）被广泛用于近似推理，并被证明对有向概率模型具有良好的性能。然而，对于传统的VAE，数据标签或特征信息是棘手的。类似地，传统的表示学习方法无法表示数据的许多突出方面。在这个项目中，我们提出了一个新的集成框架，通过结合深度度量学习来学习VAE中的潜在嵌入。通过优化VAE的平均向量上的三重损失以及VAE的标准证据下限（ELBO）来学习这些特征。这种方法，我们称之为基于三元组的变分自动编码器（TVAE），允许我们在潜在嵌入中捕获更多细粒度的信息。我们的模型在MNIST数据集上进行了测试，达到了95.60%的高三重精度，而传统的VAE（Kingma & Welling，2013）达到了75.08%的三重精度。
Deep metric learning has been demonstrated to be highly effective in learning semantic representation and encoding information that can be used to measure data similarity, by relying on the embedding learned from metric learning. At the same time, variational autoencoder (VAE) has widely been used to approximate inference and proved to have a good performance for directed probabilistic models. However, for traditional VAE, the data label or feature information are intractable. Similarly, traditional representation learning approaches fail to represent many salient aspects of the data. In this project, we propose a novel integrated framework to learn latent embedding in VAE by incorporating deep metric learning. The features are learned by optimizing a triplet loss on the mean vectors of VAE in conjunction with standard evidence lower bound (ELBO) of VAE. This approach, which we call Triplet based Variational Autoencoder (TVAE), allows us to capture more fine-grained information in the latent embedding. Our model is tested on MNIST data set and achieves a high triplet accuracy of 95.60% while the traditional VAE (Kingma & Welling, 2013) achieves triplet accuracy of 75.08%.