{"componentChunkName":"component---src-pages-research-js","path":"/research/","result":{"data":{"site":{"siteMetadata":{"title":"Kunjal Panchal"}},"allMarkdownRemark":{"edges":[{"node":{"excerpt":"","fields":{"slug":"/research-5/"},"frontmatter":{"date":"April 10, 2026","title":"Atom: Efficient On-Device Video-Language Pipelines Through Modular Reuse","tags":["On-Device Machine Learning","Mobile AI Inference","Video Retrieval","Video Assembly"],"authors":["Kunjal Panchal","Saayan Mitra","Somdeb Sarkhel","Haoliang Wang","Ishita Dasgupta","Gang Wu","Hui Guan"],"venue":"Published @ MMSys, 2026","period":"April 2026","url":"https://dl.acm.org/doi/pdf/10.1145/3793853.3795759","description":"Atom is an on-device system that speeds up video-language pipelines by decomposing large models into reusable modules, enabling parallel execution and reducing latency by ~30% with minimal performance loss."}}},{"node":{"excerpt":"","fields":{"slug":"/research-1/"},"frontmatter":{"date":"July 16, 2024","title":"Thinking Forward: Memory-Efficient Federated Finetuning of Language Models","tags":["Federated Learning","Automatic Differentiation","Forward-mode AD","First-order Gradients"],"authors":["Kunjal Panchal","Nisarg Parikh","Sunav Choudhary","Lijun Zhang","Yuriy Brun","Hui Guan"],"venue":"Published @ NeurIPS, 2024","period":"September 2024","url":"https://arxiv.org/abs/2405.15551","description":" Spry is a federated learning algorithm that enables finetuning LLMs using Forward-mode Auto Differentiation; to achieve low memory footprint, high accuracy, and fast convergence. "}}},{"node":{"excerpt":"","fields":{"slug":"/research-3/"},"frontmatter":{"date":"June 18, 2023","title":"Flow: Fine-grained Personalized Federated Learning through Dynamic Routing","tags":["Federated Learning","Personalization","Dynamic Routing"],"authors":["Kunjal Panchal","Sunav Choudhary","Nisarg Parikh","Lijun Zhang","Hui Guan"],"venue":"Published @ NeurIPS, 2023; Preliminary Presentation @ CrossFL, MLSys 2022","period":"December 2023","url":"https://arxiv.org/pdf/2211.15281.pdf","description":"Flow addresses the challenge of statistical heterogeneity in federated learning through creating a dynamic personalized model for each input instance through a routing mechanism."}}},{"node":{"excerpt":"","fields":{"slug":"/research-2/"},"frontmatter":{"date":"July 16, 2022","title":"Flash: Concept Drift Adaptation in Federated Learning","tags":["Federated Learning","Concept Drift","Drift Adaptation","Adaptive Optimization"],"authors":["Kunjal Panchal","Sunav Choudhary","Koyel Mukherjee","Subrata Mitra","Somdeb Sarkhel","Saayan Mitra","Hui Guan"],"venue":"Published @ ICML, 2023","period":"July 2023","url":"http://proceedings.mlr.press/v202/panchal23a/panchal23a.pdf","description":"Flash uses client-side early-stopping training to facilitate detection of concept drifts and the server-side drift-aware adaptive optimization in a Federated Learning setup, to solve concept drift detection and adaptation challenges.  "}}},{"node":{"excerpt":"","fields":{"slug":"/research-4/"},"frontmatter":{"date":"July 16, 2021","title":"CommunityBots: Creating and Evaluating A Multi-Agent Chatbot Platform for Public Input Elicitation","tags":["Multi-agent Chatbots","Turn-taking","Public input elicitation"],"authors":["Zhiqiu Jiang","Mashrur Rashik","Kunjal Panchal","Mahmood Jasim","Ali Sarvghad","Pari Riahi","Erica Dewitt","Fey Thurber","Narges Mahyar"],"venue":"Published @ ACM CSCW, 2023","period":"April 2023","url":"https://dl.acm.org/doi/10.1145/3579469","description":"Multi-agent chatbots, like CommunityBots, improve user engagement and response quality over single-agent systems in multi-topic conversations by effectively managing topic and agent transitions. "}}}]}},"pageContext":{}},"staticQueryHashes":["3649515864","63159454"]}