Trustworthy and Verifiable AI

Uncertainty-aware models, out-of-distribution detection, and verified decision chains for high-stakes manufacturing decisions.

Research Focus

The lab develops AI methods that can explain what they know, flag what they do not know, and support decisions with quantified confidence. This includes uncertainty quantification, coverage guarantees, out-of-distribution detection, and verification for decision pipelines that operate in manufacturing environments.

Applications

  • Manufacturing quality inspection with few-shot and multimodal models.
  • Risk-aware decision chains for smart manufacturing systems.
  • Ethical and accountable design generation for digital fabrication.

Impact

Trustworthy AI makes advanced manufacturing systems safer to deploy, easier to audit, and more reliable when data shifts, sensors degrade, or new product designs enter production.