Tanveer Syeda-Mahmood is an IBM Fellow at IBM Research and an Adjunct Professor in the Department of Biomedical Data Science at Stanford University. As a global AI leader, she is involved in many roles ranging from fundamental research, strategy, thought leadership, and management to academic/professional activities. Her research over the last three decades has spanned the field of multimodal AI covering both general AI for unstructured data management to domain-specific high precision AI in healthcare. The pioneering research and early innovations led by Tanveer and her team, have helped launch several new fields of research in AI ranging from content-based image and video retrieval in the 90s to radiology AI clinical decision support in the last decade. It resulted in over 300 refereed publications (including 10 Best Paper Awards), 170+ filed patents, mentions in 25+ textbooks, influenced hundreds of Ph.D. theses, and launched new workshops and conferences. Much of her research was also absorbed in many IBM and Xerox products, launching new lines of business for IBM (Watson Health Imaging), as well as large pilot deployments in hospitals and other enterprises. She recently received the IEEE EMBS Professional Career Achievement Award for outstanding technical achievement and leadership in multimodal decision support with lasting impact to academia/industry in multimodal healthcare AI.
Her research is focused on fundamental innovations in AI to solve complex problems, with much of the work drawing inspiration from the mechanisms in brain in developing computational models to drive AI. Examples of these over the years include a computational modeling of the visual attention mechanism in brain to drive object recognition in computer vision in the 1990s, to the latest bioinspired models for modeling human memory for memory prosthetics and next generation computer storage systems. The applications of the core technologies developed in her group are in various domains ranging from healthcare decision support, accelerated scientific discovery, to the building of next generation hybrid cloud data management and storage infrastructures.
Current research topics in general AI
- Multimodal RAG Search
- Compression in vector databases
- Rethinking relational and vector databases
- Semantic memory LLM
- Multimodal bioispired epsiodic memory VLMs (VLM, SemCLIP)
- Hippocampal learning mechanisms - Cross-modal Hopfield networks
Current research topics in healthcare AI
- Fact-checking of generative AI models
- Report correction using LLMs
- Multimodal fusion architectures for outcome prediction combining clinical and genomic data
- High precision AI for interventional imaging
- Digital Twins for Health