Associative Adversarial Networks
This is an incremental improvement for generative modeling in machine learning, addressing distribution assumptions in GANs.
The authors tackled the problem of generative adversarial networks (GANs) assuming uniform distributions for high-level features by proposing associative adversarial networks (AANs) that use Restricted Boltzmann Machines (RBMs) as associative memory to learn these distributions, with experiments showing potential.
We propose a higher-level associative memory for learning adversarial networks. Generative adversarial network (GAN) framework has a discriminator and a generator network. The generator (G) maps white noise (z) to data samples while the discriminator (D) maps data samples to a single scalar. To do so, G learns how to map from high-level representation space to data space, and D learns to do the opposite. We argue that higher-level representation spaces need not necessarily follow a uniform probability distribution. In this work, we use Restricted Boltzmann Machines (RBMs) as a higher-level associative memory and learn the probability distribution for the high-level features generated by D. The associative memory samples its underlying probability distribution and G learns how to map these samples to data space. The proposed associative adversarial networks (AANs) are generative models in the higher-levels of the learning, and use adversarial non-stochastic models D and G for learning the mapping between data and higher-level representation spaces. Experiments show the potential of the proposed networks.