A neurosymbolic Approach with Epistemic Deep Learning for Hierarchical Image Classification

In three linesNeurosymbolic framework combining Swin Transformers, focal set reasoning and differentiable fuzzy logic for hierarchical image classification. Captures epistemic uncertainty through focal sets in embedding space and enforces logical constraints via fuzzy membership functions to ensure consistency between fine and coarse predictions.

Summary generated by Claude — human-verified