Search papers, labs, and topics across Lattice.
This paper introduces a cycle-consistent neural architecture that generates natural language explanations for formal verification certificates, addressing the opacity of these certificates for non-specialist stakeholders. By employing a forward network (NN1) to map certificates to explanations and an inverse network (NN2) to reconstruct certificates, the authors achieve a remarkable 90.0% cycle-verified soundness, significantly outperforming a multi-LLM baseline. The proposed method not only enhances interpretability but also offers substantial efficiency improvements, achieving 860x faster inference times compared to traditional approaches.
Cycle-consistent neural architectures can generate clear explanations for formal verification certificates, achieving 90% soundness while being 860 times faster than traditional multi-LLM methods.
Formal verification produces machine-checkable certificates that attest to the satisfaction or violation of temporal properties, yet these certificates remain opaque to non-specialist stakeholders. We propose a cycle-consistent neural architecture that generates faithful natural language explanations of verification certificates. A forward network NN1 maps certificates to explanations, and an inverse network NN2 reconstructs certificates from explanations; a symbolic verifier closes the loop, providing a differentiable faithfulness proxy. A pointer-generator mechanism ensures lexical grounding by copying state names directly from the certificate. We evaluate on 420 test certificates spanning six verification methods (bounded proof, k-induction, inductive invariant, lasso, reachability, witness pair) in both YES and NO verdict variants, drawn from a financial compliance domain with 207 named states. Our trained architecture, combined with a hybrid inference-time routing strategy, achieves 90.0% cycle-verified soundness, surpassing a multi- LLM few-shot baseline (76.1% for the best of 16 LLM combinations across four frontier models) by 13.9 percentage points. The neural model wins on 10 of 12 verdict/kind categories, with three categories reaching 100% soundness. The architecture offers 860x faster inference (185 ms vs. 160 s per certificate for the full multi-LLM baseline), offline operation, deterministic outputs, and zero per-inference cost. These results demonstrate that trained specialization outperforms general-purpose LLM prompting for structured certificate explanation, while eliminating the deployment constraints of cloud-based inference.
