pith. sign in

arxiv: 2303.01158 · v1 · pith:KWEW6GN3new · submitted 2023-03-02 · 💻 cs.LG · cs.LO

Iterative Circuit Repair Against Formal Specifications

classification 💻 cs.LG cs.LO
keywords formalspecificationscircuitcircuitsspecificationgivenimproveslearning
0
0 comments X
read the original abstract

We present a deep learning approach for repairing sequential circuits against formal specifications given in linear-time temporal logic (LTL). Given a defective circuit and its formal specification, we train Transformer models to output circuits that satisfy the corresponding specification. We propose a separated hierarchical Transformer for multimodal representation learning of the formal specification and the circuit. We introduce a data generation algorithm that enables generalization to more complex specifications and out-of-distribution datasets. In addition, our proposed repair mechanism significantly improves the automated synthesis of circuits from LTL specifications with Transformers. It improves the state-of-the-art by $6.8$ percentage points on held-out instances and $11.8$ percentage points on an out-of-distribution dataset from the annual reactive synthesis competition.

This paper has not been read by Pith yet.

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Forward citations

Cited by 2 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score.

  1. Translating Natural Language to Strategic Temporal Specifications via LLMs

    cs.MA 2026-06 unverdicted novelty 6.0

    Framework using LLMs to translate natural language to ATL/ATL* formulas, with a new expert-validated dataset where fine-tuned 3-7B open models reach 0.84 semantic accuracy matching proprietary few-shot baselines.

  2. Translating Natural Language to Strategic Temporal Specifications via LLMs

    cs.MA 2026-06 unverdicted novelty 6.0

    The authors create a novel dataset and show that fine-tuned small open-weight LLMs can translate natural language to ATL/ATL* specifications with semantic accuracy (0.84) statistically matching strong few-shot proprie...