Balaji is a Principal Scientist at Adobe Research, India. His current research interests span the areas of natural language generation and multimodal content synthesis towards automating enterprise authoring workflows. In the past, he has worked on problems in text mining, social data analytics, high performance computing, scalable machine learning and speaker recognition.
Multimodal Content Synthesis: The experience that is consumed in various digital media involves multiple content modalities. In this line of work, the key focus is on understanding the impact of different modalities in conveying a target information and utilizing the understanding to synthesize appropriate multimodal content combination to deliver the right experience for the target medium and user need. While existing research are often limited to a single modality, this research aims to look at the combinations of different modalities and the challenges that ensues.
Tailored Natural Language Generation: There has been a growing interest in language modelling for several downstream application. Here, the key focus is to generate text that is tailored (both informationally and stylistically) to various enterprise needs. Existing works rely on the availability of parallel corpus for training end-to-end models for such task. However, our research tries to explore low resource scenarios to enable scaling of these models to enterprise settings where parallel data may not be readily available.
Causal Impacts of Content on Consumption: Enterprise content are often written to achieve certain business goals. Here, the focus is to get a holistic understanding of how a content’s information and style impacts its business goals. Our research explores causal modelling to bridge the content’s performances to its presentation, to empower marketers and copywriters towards creating impactful content.
Explainability & Interpretability in Generation: A key towards reliable adoption of automatic workflows is for the models to become more explainable and interpretable. While a large focus in NLP is to use attention (and/or other mid-layer representations) to be interpretable, they are limited to text classification and natural language inference tasks. Our research here focuses on exploring ways to make neural generation more explainable and interpretable.
Organizing Committee: Proceedings Co-Chair: 8th ACM IKDD Conference on Data Science, 2021; Tutorial Track Co-Chair: 7th ACM IKDD Conference on Data Science, 2020; Industry Track Co-Chair: 4th ACM IKDD Conference on Data Science, 2017; Demo & Exhibits Co-Chair: 9th International Conference on COMmunication Systems & NETworkS (COMSNETS), 2017; Data Challenge Co-Chair: 3rd ACM IKDD Conference on Data Science, 2016; Proceedings Co-Chair: 2nd ACM IKDD Conference on Data Science, 2015
Reviewer & Program Committee Member: Part of PC at multiple conferences in space of Computational Linguistics, Natural Language Processing, Machine Learning and Artificial Intelligence including AAAI, IJCAI, ACL, EMNLP, NAACL, INLG. Reviewed papers at multiple Journals including IEEE Transactions on Language Speech and Audio Processing, , IEEE Transactions on Neural Networks, IEEE Transactions on Information Forensics and Security, International Journal of High Performance Computing, Journal of Ambient Intelligence and Humanized Computing
Education & Experience
Balaji completed his Ph.D. in computer science at the University of Maryland in September 2011. His thesis was on Scalable Learning Methods for Speaker Recognition and Geostatistics. He completed his M.S. in electrical engineering from University of Maryland in 2008 and B.E. in electrical engineering from Anna University (India) in 2006.
His work experience includes research internships at National Institutes of Health, Bethesda, MD (May – Aug 2007) and Xerox Research Center, Webster, NY (May – Aug 2011).