Conference & Journal · 2026Open Access
Verifiable State Space Models for Deterministic Sequence Reasoning
IEEE Transactions on Pattern Analysis and Machine Intelligence (TPAMI) Vol. 48(4), pp. 1120-1135(2026). DOI: 10.1109/TPAMI.2026.3382910
IEEE Transactions on Pattern Analysis and Machine Intelligence (TPAMI)
Verifiable State Space Models for Deterministic Sequence Reasoning
Scholarly publication record and open access repository entry.
Authors & Co-Investigators
- Janusz Kowalczyk
- Aleksander Zieliński
- Karolina Nowak
Affiliated Institutions & Departments
- BBQ Institute of Advanced Science & Technology
- University of Warsaw
Abstract & Research Overview
This paper presents a novel family of verifiable state space models (SSMs) that incorporate polynomial certificate constraints into continuous-time hidden state transitions. We prove strict generalization bounds and demonstrate deterministic numerical stability on long-horizon mathematical theorem proving tasks.
Core research publication from the Laboratory for Machine Cognition & Neural Systems (MCNC Lab).
Open Access & DOI Link
Open Access article distributed under the terms of the Creative Commons Attribution 4.0 International License (CC BY 4.0).
Official publication DOI link: https://doi.org/10.1109/TPAMI.2026.3382910