Multi-Agent Diverse Generative Adversarial Networks

Despite the success of GANs in imaging, one of its major drawbacks is the problem of 'mode collapse,' where the generator learns to produce samples with extremely low variety.

To address this issue, today's guests Arnab Ghosh and Viveka Kulharia proposed two different extensions. The first involves tweaking the generator's objective function with a diversity enforcing term that would assess similarities between the different samples generated by different generators. The second comprises modifying the discriminator objective function, pushing generations corresponding to different generators towards different identifiable modes.

Guests

Arnab Ghosh: I am a second year Dphil student in Computer Vision at the Department of Engineering Science working under the supervision of Prof. Philip Torr where I work closely with Dr. Puneet Dokania. I am lucky enough to be supervised by some remarkable researchers from Adobe Richard Zhang ,Eli Shechtman, Oliver Wang , Alyosha Efros.

Viveka Kulharia: I work on Video GenAI at Moonvalley as a Member of Technical Staff. I completed my DPhil (PhD) in Computer Vision at the University of Oxford as part of the Torr Vision Group, supervised by Prof. Philip H. S. Torr and Dr. Puneet K. Dokania. Prior to that, I earned my undergraduate degree in Computer Science and Engineering (CSE) from the Indian Institute of Technology, Kanpur, where I was advised by Prof. Vinay Namboodiri, Prof. Amitabha Mukerjee, and Prof. Harish Karnick.