Conditioned Generative Model via Latent Semantic Controlling for Learning Deep Representation of Data

摘要

Learning representations of data is an important issue in machine learning. Though generative adversarial network has led to significant improvements in the data representations, it still has several problems such as unstable training, hidden manifold of data, and huge computational overhead. Moreover, most of GAN's have a large size of manifold, resulting in poor scalability. In this paper, we propose a novel GAN to control the latent semantic representation, called LSC-GAN, which allows us to produce desired data and learns a representation of the data efficiently. Unlike the conventional GAN models with hidden distribution of latent space, we define the distributions explicitly in advance that are trained to generate the data based on the corresponding features by inputting the latent variables, which follow the distribution, into the generative model. As the larger scale of latent space caused by deploying various distributions makes training unstable, we need to separate the process of defining the distributions explicitly and operation of generation. We prove that a variational auto-encoder is proper for the former and modify a loss function of VAE to map the data into the corresponding pre-defined latent space. The decoder, which generates the data from the associated latent space, is used as the generator of the LSC-GAN. Several experiments on the CelebA dataset are conducted to verify the usefulness of the proposed method. Besides, our model achieves a high compression ratio that can hold about 24 pixels of information in each dimension of latent space.