סמינר תואר שני

  You are invited to attend a lecture 

Join Zoom Meeting

https://tau-ac-il.zoom.us/j/89506193877?pwd=TnYvTUc5b3I0eXVYNkJINWUzSjZHdz09

COMMUTE-GAN: COMPETETITIVE, MULTIPLE, EFFICIENT GENERATIVE ADVERSARIAL NETWORKS

 By:

 Omkar Joglekar

 MSc student under the supervision of Prof. Goren Gordon

Abstract

 In complex creative scenarios, co-creativity by multiple agents offers great advantages. Each agent has a specific skill set and a set of abilities, which is generally not enough to perform a large and complex task single-handed.
These kinds of tasks benefit substantially from collaboration. In deep learn-ing applications, data generation is an example of such a complex, poten-tially multi-modal task. Previous Generative Adversarial Networks (GANs)focused on using a single generator to generate multi-modal datasets, which is sometimes known to face issues such as mode-collapse and failure to converge. Single generators also must be very large so that they can generalize com-plex datasets, so this method can easily run into memory constraints. The current multi-generator-based works such as MGAN, MMGAN, MADGAN and AdaGAN either require training a classifier online, the use of complex mixture models or sequentially adding generators, which is computationally complex. In this work, we present a simple, novel approach of training com- petitive multiple efficient GANs (ComMutE-GANs), with multiple genera- tors and a single critic/discriminator, without introducing external complex- ities such as a classifier model. We introduce a new component to the genera- tor loss during GAN training, based on the Total Variation Distance (TVD). Our method offers a robust, stable, memory efficient and easily parallelizable architecture. We present a proof-of-concept on the MNIST dataset, which has 10 modes of data. The individual generators learn to generate different digits from the distribution, and together learn to generate the whole dis- tribution. We compare ComMutE-GANs with larger single-generator GANs and show its memory efficiency and increased accuracy.