Distilling Formal Logic into Neural Spaces: A Kernel Alignment Approach for Signal Temporal Logic
A teacher-student framework that distills symbolic STL kernels into Transformer encoders via geometric alignment, bridging formal logic and neural representations.
Mar 5, 2026
Bridging Logic and Learning: Decoding Temporal Logic Embeddings via Transformers
We present a Transformer-based approach to decode STL embeddings, enabling interpretable and semantically accurate formula generation, with applications to requirement mining.
Jul 10, 2025