Large Language Models and multimodal systems introduce unique uncertainty challenges, such as uncertainty compounding during autoregressive generation and dynamic shifts in uncertainty based on context, requiring uncertainty quantification approaches tailored to their specific characteristics rather than methods for simpler discriminative models, as cited in reference [56].

Authors

• Person: Not available Organization: Springer
  A Comprehensive Review of Neuro-symbolic AI for Robustness ...

Sources

• A Comprehensive Review of Neuro-symbolic AI for Robustness ... link.springer.com Springer via serper

Referenced by nodes (2)

• Large Language Models concept
• Uncertainty quantification concept