Variational autoencoders with triplet loss for representation learning | IEEE Conference Publication | IEEE Xplore