Inferential Machine Learning: Towards Human-collaborative Foundation Models - presented by Professor Ghassan AlRegib

Inferential Machine Learning: Towards Human-collaborative Foundation Models

Professor Ghassan AlRegib

Professor Ghassan AlRegib
Computational Science Seminars
Host
Research Communities by Springer Nature
DateThursday, March 13, 2025 2:00 PM to 3:00 PM (UTC)
Live eventThe live event will be accessible via this page.
Research Communities by Springer Nature
Inferential Machine Learning: Towards Human-collaborative Foundation Models
Professor Ghassan AlRegib
Ghassan AlRegib
Georgia Institute of Technology

Neural network driven applications like ChatGPT suffer from hallucinations where they confidently provide inaccurate information. A funda- mental reason for this inaccuracy is the feed-forward nature of inductive decisions taken by neural networks. Such decisions are a result of training schemes that do not allow networks to deviate from and creatively abduce reasons at inference. With the advent of foundation models that are adapted across applications and data, humans can directly intervene and prompt vision-language foundation models. However, without understanding the operational limits of the underlying networks, human interventions often lead to unfair, inaccurate, hallucinated and unintelligible outputs. These outputs undermine the trust in foundation models, thereby causing roadblocks to their adoption in everyday lives. In this talk, we review systematic ways to analyze and understand human interventions in neural network functionality at inference. Specifically, we show that a human-AI collaborative environment via inferential machine learning techniques is a promising endeavor.

References
  • 1.
    Ryan Benkert et al. (2024) Effective Data Selection for Seismic Interpretation through Disagreement.
  • 2.
    Mohit Prabhushankar and Ghassan AlRegib (2024) VOICE: Variance of Induced Contrastive Explanations to quantify Uncertainty in Neural Network Interpretability.
  • 3.
    Kiran Kokilepersaud et al. (2024) HEX: Hierarchical Emergence Exploitation in Self-Supervised Algorithms.
  • 4.
    Jorge Quesada et al. (2024) Benchmarking Human and Automated Prompting in the Segment Anything Model.
Date & time
Mar
13
2025
Thursday, March 13, 2025 2:00 PM to 3:00 PM (UTC)
Details
Listed seminar This seminar is open to all
Recorded Available to all
Q&A Open on this page for 1 day after the seminar
Disclaimer The views expressed in this seminar are those of the speaker and not necessarily those of the journal