Challenges & Opportunities for Real World Applications
of Foundation Models
Large Language Models (LLMs) Workshop
Open to Lehigh University faculty, post-docs and graduate students.
AI transformation, large language models (LLMs) and other foundation models have become central to diverse applications, spanning natural language processing to computer vision. These models, with their immense potential, are revolutionizing research and society but their application in real-world settings poses some challenges. This workshop explores the lessons learnt from deployment of such models in the domains of robotics, human-centric applications, cyber physical systems, healthcare and finance, as well as discussing techniques that have been explored to adapt foundation models so that they can be ready for real deployments in these domains.
Large Language Models (LLMs) Research at Lehigh
I-DISC's LLM Research Group consists of 4 subgroups focusing on different aspects of Large Language Models (LLMs) / Vision Language Models (VLMs) related research:
Foundation Models Advancement (FMA)
The FMA subgroup focuses on addressing the limitations of existing LLMs/VLMs, creating justifications for LLM results, and improving security & privacy of LLM/VLMs.
Sub group Leader: Jeff Heflin, Computer Science and Engineering, RCEAS
Human-centric LLMs (HCL)
The HCL subgroup focuses on understanding how well LLMs/VLMs mimic human behaviors, how to address potential biases in LLM/VLM when they are being used to mimic human behaviors, how to mitigate against misinformation issues that are generated by LLMs/VLMs for example.
Sub group Leader: Rebecca Wang, Marketing, College of Business
LLMs for CPS / Robotics (LCPS)
The LCPS subgroup focuses on how to apply LLMs/VLMs to cyber physical systems and robots with formal guarantees.
Sub group Leader: Parv Venkitasubramaniam, Electrical and Computer Engineering, RCEAS
LLMs for Healthcare (LLH)
The LLH subgroup focuses on how to responsibly apply LLMs/VLMs to health care domains. This includes ensuring such models are fair, scalable and accurate.
Sub group Leader: Mooi Choo Chuah, Computer Science and Engineering, RCEAS
Building C, Mountaintop Campus, Lehigh University | Room 210