We work at the intersection of ML Research and ML Engineering. To a great extent, it is hard to decouple these two disciplines. ML, at its heart, is an empirical discipline. The theoretical foundations of ML are sound in certain areas and evolving in other areas as we continue to push the envelope.The projects we work on involve natural language processing, computational linguistics, large language models (LLMs), networking, security, knowledge acquisition, insight generation, and related areas. We firmly believe in the “Farm to Table” approach to ML: where possible, we curate our own data to train our own models, and productize them through our unique ML operations (MLOps) approach for better customer experience. As part of this process, we gather new insights which we publish at peer-reviewed conferences and disseminate as technical talks.
How susceptible are LLMs to logical fallacies?
Amirreza Payandeh, Dan Pluth, Jordan Hosier, Xuesu Xiao , Vijay K. Gurbani, accepted for publication in The 2024 Joint International Conference on Computational Linguistics, Language Resources and Evaluation (LREC-COLING 2024), Torino, Italy.
Evaluating feature importance for speaker separation models
Deeksha Prabhaker, Jose Ignacio Pozuelo, Daniel Pluth , Ayush Panda and Vijay K. Gurbani, accepted for publication in the IEEE SoutheastCon 2024, Atlanta, GA.
Digits micro-model for accurate and secure transactions
Chirag Chhablani, Nikhita Sharma, Jordan Hosier, and Vijay K. Gurbani. (To be published) AAAI 2024.
A low latency technique for speaker detection from a large negative list
Yu Zhou, Vijay K. Gurbani, B. Chandra Mouli. ICNLSP 2022: 202-211
Improved Named Entity Recognition for Noisy Call Center Transcripts
Sam Davidson, Jordan Hosier, Yu Zhou, Vijay K. Gurbani. W-NUT 2021: 361-370
Lightweight domain adaptation: A filtering pipeline to improve accuracy of an Automatic Speech Recognition (ASR) engine
Jordan Hosier, Yu Zhou, Nikita Sharma, Vijay K. Gurbani. ACAI 2021: 95:1-95:9
Project Vac: Can a Text-to-Speech Engine Generate Human Sentiments?
Shivam Kulkarni, Luis Barbado, Jordan Hosier, Yu Zhou, Siddharth Rajagopalan, and Vijay K. Gurbani. SpeD 2021: 103-108
Semantic Search Pipeline: From Query Expansion to Concept Forging
Elizabeth Soper, Jordan Hosier, Dustin Bales, and Vijay K. Gurbani. ICDE 2021: 2309-2314
A systematic study of open source and commercial text-to-speech (TTS) engines
Jordan Hosier, Jordan Kalfen, Nikita Sharma, and Vijay K. Gurbani. TDS 2020: 312-320
Disambiguation and Error Resolution in Call Transcripts
Jordan Hosier, Vijay K. Gurbani, Neil Milstead. IEEE BigData 2019: 4602-4607
Low latency technique for speaker identification, Yu Zhou, Vijay K. Gurbani, and Bopsi Chandramouli, US Patent Application US 2023-0386476-A1, Publication date 2023-05-23, available online from https://ppubs.uspto.gov/dirsearch-public/print/downloadPdf/20230386476
Conversational interface and method for interaction with communication devices, Nikhita Sharma, Joseph R. Smetana, David J. Fruin, and Vijay K. Gurbani, US Patent Application US-20230127163-A1, Publication date 2023-04-27, available online from https://image-ppubs.uspto.gov/dirsearch-public/print/downloadPdf/20230127163
System and method for disambiguation and error resolution in call transcripts, Vijay K. Gurbani, Jordan Hosier, Yu Zhou, Nikhita Sharma, and Neil Milstead, US Patent Application US-20210142789-A1, Publication date 2021-05-13, available online from https://image-ppubs.uspto.gov/dirsearch-public/print/downloadPdf/20210142789