Vignesh Gokul

Hi! I am a Ph.D. candidate in Computer Science at UC San Diego advised by Prof. Shlomo Dubnov. My research interests includes Creative Interaction/Improvisation, Video/Audio Understanding and Federated Learning.

I am on the job market for research roles in the industry (exp. graduation by June 2023)

Email  /  Google Scholar  /  Github

profile photo
Research Sumary

  • Co-creativity/Improvisation: My research goal is to enable creative interactions between humans and visual assistants. These interactions can be either visual (dance) or audio (music improvisation) .

  • Distributed Co-creativity: I also focus on building federated/distributed systems that can learn such interactions in the absence of large datasets.

Switching Machine Improvisation Models by Latent Transfer Entropy Criteria
Shlomo Dubnov, Vignesh Gokul, Gerard Assayag
International Conference on Bayesian and Maximum Entropy methods in Science and Engineering, 2022

We introduce Symmetric Transfer Entropy (SymTE) as a quantitative metric to switch between generative models based on a control signal.

Creative Improvised Interaction with Generative Musical Systems
Shlomo Dubnov, Gerard Assayag, Vignesh Gokul
IEEE 5th International Conference on Multimedia Information Processing and Retrieval (MIPR), 2022

In this paper we survey the methods for control and cre-ative interaction with pre-trained generative models for audio and music.

FedLTN: Federated Learning for Sparse and Personalized Lottery Ticket Networks
Vaikkunth Mugunthan*, Eric Lin,*, Vignesh Gokul, Christian Lau, Lalana Kagal, Steve Pieper
ECCV, 2022 (Poster)

In this paper, we propose FedLTN, a novel approach motivated by the well-known Lottery Ticket Hypothesis to learn sparse and personalized lottery ticket networks (LTNs) for communication-efficient and person- alized FL under non-identically and independently distributed (non-IID) data settings.

Bias-Free FedGAN: A Federated Approach to Generate Bias-Free Datasets
Vignesh Gokul*, Vaikkunth Mugunthan*, Lalana Kagal, Shlomo Dubnov
Preprint, 2022

Federated GANs propagate data biases onto the global model. We propose Bias-Free FedGAN, an approach to generate bias-free synthetic datasets using FedGAN.

DPD-InfoGAN: Differentially Private Distributed InfoGAN
Vaikkunth Mugunthan*,Vignesh Gokul*, Lalana Kagal, Shlomo Dubnov
EuroMLSys, 2021

We propose a differentially private version of InfoGAN (DP-InfoGAN). We also extend our framework to a distributed setting (DPD-InfoGAN) to allow clients to learn different attributes present in other clients' datasets in a privacy-preserving manner.

Semantic Interaction with Human Motion Using Query-Based Recombinant Video Synthesis
Vignesh Gokul, Ganesh Prasanna Balakrishnan, Tammuz Dubnov, Shlomo Dubnov

In this paper we describe a gestural motif extraction system that combines deep feature learning with structural similarity analysis to allow such query based human-computer motion interaction.

Website template from Jon Barron