mlsanityData to 7 Oct 2026

DeepDFA: Injecting Temporal Logic in Deep Learning for Sequential Subsymbolic Applications

arXiv 2602.03486 ↗First posted 3 Feb 2026cs.LG cs.AIcited by 0

Not on the map. Why?

Abstract (arXiv)

Integrating logical knowledge into deep neural network training is still a hard challenge, especially for sequential or temporally extended domains involving subsymbolic observations. To address this problem, we propose DeepDFA, a neurosymbolic framework that integrates high-level temporal logic - expressed as Deterministic Finite Automata (DFA) or Moore Machines - into neural architectures. DeepDFA models temporal rules as continuous, differentiable layers, enabling symbolic knowledge injection into subsymbolic domains. We demonstrate how DeepDFA can be used in two key settings: (i) static image sequence classification, and (ii) policy learning in interactive non-Markovian environments. Across extensive experiments, DeepDFA outperforms traditional deep learning models (e.g., LSTMs, GRUs, Transformers) and novel neuro-symbolic systems, achieving state-of-the-art results in temporal knowledge integration. These results highlight the potential of DeepDFA to bridge subsymbolic learning and symbolic reasoning in sequential tasks.

Key facts. No citing papers in the map yet (data through 7 Oct 2026). Builds on 34 of its 62 references that are in the map. First posted on arXiv 3 Feb 2026.

Builds on

34 of 62 references.

References · 34

  1. 2016Asynchronous Methods for Deep Reinforcement Learningcited by 2,218
  2. 2019Neural Probabilistic Logic Programming in DeepProbLogcited by 389
  3. 2017A Semantic Loss Function for Deep Learning with Symbolic Knowledgecited by 267
  4. 2020Logic Tensor Networkscited by 197
  5. 2020Reward Machines: Exploiting Reward Function Structure in Reinforcement Learningcited by 183
  6. 2020Graph Neural Networks Meet Neural-Symbolic Computing: A Survey and Perspectivecited by 93
  7. 2023NeurASP: Embracing Neural Networks into Answer Set Programmingcited by 92
  8. 2021LTL2Action: Generalizing LTL Instructions for Multi-Task RLcited by 72
  9. 2018Foundations for Restraining Bolts: Reinforcement Learning with LTLf/LDLf restraining specificationscited by 69
  10. 2017Environment-Independent Task Specifications via GLTLcited by 67
  11. 2021DeepStochLog: Neural Stochastic Logic Programmingcited by 58
  12. 2019Reinforcement Learning with Non-Markovian Rewardscited by 45
22 more
  1. 2020Neural-Symbolic Integration: A Compositional Perspectivecited by 45
  2. 2021Modular Deep Reinforcement Learning for Continuous Motion Planning with Temporal Logiccited by 38
  3. 2020Encoding formulas as deep networks: Reinforcement learning for zero-shot execution of LTL formulascited by 36
  4. 2022ROAD-R: The Autonomous Driving Dataset with Logical Requirementscited by 34
  5. 2020Induction and Exploitation of Subgoal Automata for Reinforcement Learningcited by 26
  6. 2020Active Finite Reward Automaton Inference and Reinforcement Learning Using Queries and Counterexamplescited by 23
  7. 2020Reinforcement Learning Based Temporal Logic Control with Maximum Probabilistic Satisfactioncited by 22
  8. 2017Symbolic LTLf Synthesiscited by 21
  9. 2019Specification Patterns for Robotic Missionscited by 16
  10. 2019Hybrid Compositional Reasoning for Reactive Synthesis from Finite-Horizon Specificationscited by 16
  11. 2022Joint Learning of Reward Machines and Policies in Environments with Partially Known Semanticscited by 12
  12. 2024Reward Machines for Deep RL in Noisy and Uncertain Environmentscited by 11
  13. 2022Markov Abstractions for PAC Reinforcement Learning in Non-Markov Decision Processescited by 10
  14. 2022Noisy Symbolic Abstractions for Deep RL: A case study with Reward Machinescited by 10
  15. 2024NeSyA: Neurosymbolic Automatacited by 9
  16. 2024DeepDFA: Automata Learning through Neural Probabilistic Relaxationscited by 8
  17. 2025Reinforcement Learning with Stochastic Reward Machinescited by 7
  18. 2024Neural Reward Machinescited by 6
  19. 2024Detecting Hidden Triggers: Mapping Non-Markov Reward Functions to Markovcited by 5
  20. 2024Conformance Checking of Fuzzy Logs against Declarative Temporal Specificationscited by 5
  21. 2021Monitoring Hybrid Process Specifications with Conflict Management: The Automata-theoretic Approachcited by 4
  22. 2024Play Everywhere: A Temporal Logic based Game Environment Independent Approach for Playing Soccer with Robotscited by 2

Built on by

Not cited yet.

Similar, not cited

Close in wording, no citation either way

  1. 2026Time to Reason: Scalable Neurosymbolic Learning for LTLf via Fuzzy Semantics
  2. 2021Embedding Symbolic Temporal Knowledge into Deep Sequential Models
  3. 2025GradSTL: Comprehensive Signal Temporal Logic for Neurosymbolic Reasoning and Learning
  4. 2025Reasoning in Neurosymbolic AI
  5. 2025Enhancing Symbolic Machine Learning by Subsymbolic Representations
  6. 2025LTLZinc: a Benchmarking Framework for Continual Learning and Neuro-Symbolic Temporal Reasoning
  7. 2023A Novel Neural-symbolic System under Statistical Relational Learning
  8. 2024Symbol Correctness in Deep Neural Networks Containing Symbolic Layers