TVAE: Triplet-Based Variational Autoencoder using Metric Learning
This work addresses the limitation of traditional VAEs in capturing fine-grained data features, which is important for applications requiring detailed similarity measures, but it is incremental as it combines existing techniques.
The paper tackles the problem of learning more informative latent embeddings in variational autoencoders (VAEs) by integrating deep metric learning with triplet loss, achieving a triplet accuracy of 95.60% on MNIST compared to 75.08% for traditional VAE.
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%.