Pith. sign in

REVIEW 3 major objections 5 minor 1 cited by

Neuro-Symbolic Predictive Process Monitoring

T0 review · 3 major / 5 minor · reviewed 2026-08-05 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

Pith's one-line read The paper claims that adding a differentiable logic loss to autoregressive suffix predictors improves both edit-distance accuracy and satisfaction of temporal-logic constraints.

desk verdict The local loss is a clean idea, but the main empirical claim is built into the protocol: LTLf constraints are mined from the test set and the test set is filtered to satisfy them. read the letter →

arxiv 2509.00834 v2 pith:3AP7UXWL submitted 2025-08-31 cs.AI cs.FLcs.LGcs.LO

Axel Mezini , Elena Umili , Ivan Donadello , Fabrizio Maria Maggi , Matteo Mancanelli , Fabio Patrizi This is my paper · ORCID
classification cs.AIcs.FLcs.LGcs.LO
keywords PredictiveprocessmonitoringSuffixpredictionNeuro-symbolicAILTLfDifferentiableautomataLogiclossGumbel-SoftmaxDeclareconstraints
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

The paper tries to show that a neural model predicting the future portion of a business-process trace learns better when logical rules about the process are folded into training rather than checked only after generation. Its recipe adds a differentiable logic-loss term to the standard next-event prediction loss: a temporal-logic rule is translated into a finite-state automaton, embedded as a differentiable layer, and complete predicted suffixes are sampled with Gumbel-Softmax so gradients flow through both accuracy and rule compliance. On three real-world event logs, this hybrid training yields predicted suffixes closer to the true continuation by edit distance and far more likely to satisfy the rules, with compliance near 100 percent even when 40 percent of training labels are corrupted. The same recipe is offered as a general way to inject temporal knowledge into any autoregressive symbolic sequence generator.

What carries the argument

Load-bearing is the combined loss L = αLD + (1 − α)Lφ. LD is per-token cross-entropy. Lφ uses an LTLf formula translated into a deterministic finite automaton (extended with an end-of-trace marker and, for the local variant, failing states), DeepDFA — a tensor representation that makes acceptance differentiable over probabilistically grounded traces — and Gumbel-Softmax sampling, which yields near-one-hot suffixes whose gradients back-propagate. Lloc penalizes entering a failing state at each step; Lglob maximizes a Monte Carlo estimate of the fraction of sampled traces the automaton accepts.

What would settle it

Run the same comparison with the temporal rules mined only from the training split (or supplied by an expert) and keep all test traces in the evaluation, including those that violate the rules. If the logic-trained models no longer beat the data-only baseline on satisfaction or on Damerau-Levenshtein similarity, the reported improvement is an artifact of test-derived filtering.

Watch

Extended reading notes

Core claim

The paper claims that adding a differentiable logic-loss term to the usual cross-entropy loss makes an autoregressive suffix predictor produce continuations that are closer to the observed ones and far more likely to satisfy stated temporal rules. The loss translates the LTLf rule into a deterministic finite automaton, embeds it as a differentiable tensor layer, and samples complete suffixes with Gumbel-Softmax so gradients flow through sampling. A local variant penalizes moves into automaton states from which satisfaction is impossible; a global variant maximizes a Monte Carlo estimate of satisfaction. On three real-world event logs with up to 40 percent label noise, logic-trained models st

Load-bearing premise

The reported compliance gain rests on the rules being mined from the test set at 85 percent support and on traces that violate them being removed from both training and test data, so part of the satisfaction rate is built into the evaluation rather than coming from independent prior knowledge.

Editorial extensions

If this is right

  • Models trained with either logic loss satisfy the mined temporal rules at close to 100 percent even when 40 percent of training events are relabeled as noise.
  • Adding the logic loss does not reduce suffix similarity to the ground truth; reported Damerau-Levenshtein similarity stays at least as high as the data-only RNN.
  • The global logic loss consistently reaches convergence in fewer training epochs than the data-only baseline, up to roughly a third of the epochs in some configurations.
  • Because the method only needs an autoregressive next-symbol distribution, it transfers to any sequence predictor, not just the LSTM used in the experiments.
  • The local loss gives step-level feedback but requires automata with failing states; the global loss handles arbitrary formulas at higher computational cost.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • The test-derived mining protocol makes the reported compliance rates an upper bound; re-running with constraints written before seeing test traces is the direct way to test how much of the gain is genuinely prior knowledge.
  • The convergence speedup hints that the logic loss acts as a regularizer; if so, the accuracy gap between logic-trained and data-only models should widen as label noise increases and narrow on clean logs.
  • The self-loop alphabet extension becomes impractical for very large vocabularies, so transferring the recipe to large language models would require a different way to handle unconstrained symbols.
  • A combined local-plus-global loss is a natural next step for mixed safety/liveness rule sets; the authors list it as future work.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

3 major / 5 minor

Summary. The paper proposes a neuro-symbolic approach to suffix prediction in business process monitoring. It integrates LTLf constraints into the training of an autoregressive next-activity predictor by defining a differentiable logic loss, with a local variant (Lloc) that penalizes transitions into permanently failing DFA states and a global variant (Lglob) that uses Gumbel-Softmax sampling and DeepDFA to maximize the Monte Carlo estimate of suffix satisfaction. The combined loss is L = αLD + (1 − α)Lφ. The authors evaluate RNN, RNN+LLL, and RNN+GLL on three event logs under increasing noise, reporting Damerau-Levenshtein similarity and LTLf satisfaction rate. The central claim is that incorporating LTLf knowledge at training time improves both suffix similarity and constraint satisfaction. However, Section 4.1 describes a protocol in which Declare constraints are mined from the test set and traces violating them are removed from both training and test sets, which compromises the empirical evaluation.

Significance. If the empirical claim were soundly supported, the paper would make a useful contribution: a model-agnostic way to inject temporal-logic knowledge into training of symbolic sequence generators, with two loss variants that are clearly motivated and backed by public code. The conceptual machinery—DeepDFA encoding, Gumbel-Softmax differentiable sampling, and the local/global loss decomposition—is coherent and potentially reusable beyond BPM. The paper also provides a reproducibility link and transparent dataset statistics. However, the current experimental protocol does not establish the claimed benefit. Because the LTLf model is derived from the test set and the test set is filtered to satisfy it, the reported satisfaction gains are partly guaranteed by construction, and the comparison against the base RNN is biased. The evaluation needs to be redone with genuinely prior knowledge and an unfiltered test set before the central claim can be accepted.

major comments (3)
  1. [4.1 Experimental Setup] The protocol violates the 'prior knowledge' assumption that the paper's method relies on. The text states: 'Knowledge is extracted from the test set in the form of Declare constraints using the Declare Miner... traces that violate the extracted LTLf model are removed from both the training and test set.' This makes the LTLf formula a posterior summary of the test labels, not background knowledge available before data collection. It also removes non-conforming test traces, so the satisfaction-rate metric is computed only on traces that satisfy the formula by construction. The base RNN is denied access to this test-derived formula while being evaluated on the filtered set, whereas the proposed models are explicitly trained to maximize satisfaction of that same formula. Consequently, the reported SAT improvements and even the DL comparisons are confounded by test-set leakage and distributio
  2. [4.2 Empirical Results; Eq. (14)-(15)] The main success metric is circular with the training objective. The global loss is Lglob = −log(ˆPθ⊨ϕ), where ˆPθ⊨ϕ is the empirical satisfaction probability over sampled suffixes. The paper then reports the satisfaction rate of predicted traces on a test set that has been filtered to satisfy exactly the same LTLf model. Thus, SAT values near 100% are expected for the GLL method and are weak evidence that training-time knowledge injection works. Satisfaction-rate differences between methods are largely a consequence of the experimental setup. The authors should report compliance on the original unfiltered test set, and ideally with respect to independently held-out constraints, to provide meaningful evidence.
  3. [4.1-4.2 and Table 2] The paper does not report the number of traces removed from the training/test sets, nor any dispersion measures for the results of the 15 runs. Table 2 gives average epochs without standard deviations or significance tests, so the claim that knowledge 'accelerates model convergence' is not quantified. Figure 2 and Figure 3 do not show confidence intervals or error bars. Given the small number of datasets and the filtering step, these omissions make it difficult to assess whether observed differences are robust. The authors should report the fraction of filtered traces and standard errors/confidence intervals, and ideally perform statistical tests for the main comparisons.
minor comments (5)
  1. [4.1 Experimental Setup] The set of free hyperparameters (α, τ, N, Declare minimum support) is listed, but the exact values used for each dataset are not reported. This makes reproduction harder even with the linked code. Please include a table or appendix with the chosen values.
  2. [5 Related Work] The claim of being 'the first to integrate temporal knowledge in the generation of multi-step symbolic sequences at training time' is too strong given STLnet [6], which is acknowledged but dismissed on domain grounds. The statement should be qualified to discrete domains or LTLf specifically.
  3. [3.2 Local Guidance] The paper correctly notes that Lloc only penalizes permanent violations (case (i)) and not other non-satisfying traces (case (ii)). This is a real limitation; the authors should discuss how often case (ii) occurs in the experiments, especially since the DFA failure states are few (Table 1).
  4. [4.2 Empirical Results] Figures 2 and 3 are referenced before they appear in the text; please reorder or fix the cross-reference. Also, the figures would benefit from error bars or shaded confidence regions.
  5. [General] Minor typos and notation inconsistencies exist, e.g., 'subsymbolic' (Section 2.1) and the use of both '˜σ' and 'σ' in Eq. (8). A careful proofreading pass is recommended.

Circularity Check

1 steps flagged · score 6.0 of 10

Central empirical evidence is partly circular: LTLf constraints are mined from the test set and violating test traces are removed, so the 'prior' knowledge is a posterior test-set summary and the satisfaction metric is measured on a filtered set guaranteed to satisfy it.

  1. fitted input called prediction [Section 4.1, Experimental Setup, first paragraph]
    "Knowledge is extracted from the test set in the form of Declare constraints using the Declare Miner [38], each with a minimum support value of 85% (i.e., satisfied in at least 85% of the traces). These constraints are then translated into their corresponding LTLf formulas and combined in an LTLf model using conjunctions. To enable controlled experiments, traces that violate the extracted LTLf model (i.e., that do not satisfy all the discovered constraints) are removed from both the training and test set."

    The Declare constraints are mined from the test set (Declare Miner, 85% support), so the 'prior' LTLf model is a posterior summary of the test labels, not independent background knowledge. Then all test traces violating this extracted model are removed from both training and test sets. Therefore, by construction, every remaining test trace satisfies the LTLf model, and the reported satisfaction-rate metric is evaluated against that same test-derived formula on a filtered set that was defined to be compliant with it. The SAT result is thus partly guaranteed by the protocol—not solely by the claimed training-time benefit—and the base RNN is denied access to the test-derived formula. The paper does not report how many traces were removed, so the bias cannot be assessed.

full rationale

The derivation of the two logic losses (Sections 3.2–3.5) is internally non-circular: L_loc is defined from DFA failing states, L_glob is a Monte Carlo estimate of Pθ⊨ϕ, and both are combined with the supervised loss αLD+(1−α)Lϕ; the components (ltlf2dfa, DeepDFA, Gumbel-Softmax) are standard and independently published. No load-bearing self-citation chain or uniqueness argument is used; the self-citations to prior work [17,26,48,49] are tool/metadata references, not premises that assume the conclusion. The circularity is confined to the evaluation protocol. Section 4.1 mines the LTLf constraints from the test set and then deletes non-compliant test traces, making the 'prior knowledge' posterior and making the filtered test set satisfy the formula by construction. The satisfaction-rate metric is therefore measured against a target derived from the same data and on a subset selected to meet that target. This partially reduces the central empirical claim to the experimental setup. The DL-similarity results are not forced in the same direct way, so the paper is not wholly circular; however, the main evidence for constraint-satisfaction improvement is compromised, warranting a score of 6.

Assumptions & free parameters 4 free parameters · 6 assumptions · 0 invented entities

The central method relies on the standard LTLf-to-DFA translation and on the authors' DeepDFA relaxation; the main non-standard assumption is treating test-mined Declare constraints as prior knowledge, which drives the experimental circularity.

free parameters (4)
  • alpha (logic loss weight) = not reported
    In Eq. 11 the total loss is alpha * LD + (1 - alpha) * L_phi; no value, tuning range, or sensitivity analysis is given, so all results depend on an unspecified choice.
  • tau (Gumbel-Softmax temperature) = not reported
    In Eq. 19 tau controls the sharpness of sampled suffixes and affects both gradient quality and the compliance measured by DeepDFA; it is not specified.
  • N (Monte Carlo samples for global loss) = not reported
    Eq. 14 averages over N sampled suffixes; the variance and bias of the estimated satisfaction probability depend on N, which is never stated.
  • Declare minimum support threshold = 85%
    Section 4.1: constraints are mined with minimum support 85%; changing this threshold changes the LTLf model, the filtered data, and all reported metrics.
assumptions (6)
  • standard math LTLf to DFA translation via ltlf2dfa is correct and equivalent (Eq. 3).
    Invoked in Section 3.4.1 to build the automaton; relies on the cited translation tool [28] and known LTLf-to-DFA results.
  • domain assumption Mutual exclusivity of activities: exactly one activity is true at each trace position.
    Assumed after Eq. 1 as customary in BPM; it keeps the DFA alphabet at A instead of 2^A.
  • domain assumption DeepDFA's probabilistic matrix relaxation approximates the deterministic DFA acceptance function with usable gradients.
    Sections 3.3 and 3.4.2 use DeepDFA [26] as the differentiable evaluator; no theorem or experiment in this paper validates that soft states track discrete acceptance.
  • domain assumption Gumbel-Softmax samples at the chosen tau are close enough to one-hot for DeepDFA compliance to be meaningful and differentiable.
    Section 3.5, Eq. 19; tau is unreported, so the approximation regime is unknown.
  • ad hoc to paper Declare constraints mined from the test set with 85% support are treated as reliable prior knowledge.
    Section 4.1: the paper calls this background knowledge, but it is data-mined from the evaluation set, which makes the prior knowledge depend on the test data.
  • domain assumption The first EOT symbol ends the evaluated trace and symbols after the first EOT are irrelevant.
    Section 3.4.1 introduces terminal states and defines compliance only at first EOT; this choice defines the satisfaction metric.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Neuro-Symbolic Predictive Process Monitoring." pith.science (2026). https://pith.science/paper/3AP7UXWL

@misc{pith2026250900834,
  author       = {Pith},
  title        = {Pith review of: Neuro-Symbolic Predictive Process Monitoring},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/3AP7UXWL}},
  note         = {Machine review of arXiv:2509.00834}
}
read the original abstract

This paper addresses the problem of suffix prediction in Business Process Management (BPM) by proposing a Neuro-Symbolic Predictive Process Monitoring (PPM) approach that integrates data-driven learning with temporal logic-based prior knowledge. While recent approaches leverage deep learning models for suffix prediction, they often fail to satisfy even basic logical constraints due to the lack of explicit integration of domain knowledge during training. We propose a novel method to incorporate Linear Temporal Logic over finite traces (LTLf) into the training process of autoregressive sequence predictors. Our approach introduces a differentiable logical loss function, defined using a soft approximation of LTLf semantics and the Gumbel-Softmax trick, which can be combined with standard predictive losses. This ensures that the model learns to generate suffixes that are both accurate and logically consistent. Experimental evaluation on three real-world datasets shows that our method improves suffix prediction accuracy and compliance with temporal constraints. We also introduce two variants of the logic loss (local and global) and demonstrate their effectiveness under noisy and realistic settings. While developed in the context of BPM, our framework is applicable to any symbolic sequence generation task and contributes to advancing Neuro-Symbolic AI.

Figures

Figures reproduced from arXiv: 2509.00834 by the authors.

Figure 1
Figure 1. Global logic loss computation using a differentiable procedure for both suffix [PITH_FULL_IMAGE:figures/full_fig_p014_1.png] view at source ↗
Figure 2
Figure 2. Comparison of satisfiability (SAT) of predicted traces with respect to logical [PITH_FULL_IMAGE:figures/full_fig_p017_2.png] view at source ↗
Figure 3
Figure 3. Comparison of predicted suffix similarity (based on scaled Damerau-Levenshtein [PITH_FULL_IMAGE:figures/full_fig_p018_3.png] view at source ↗

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 1 Pith paper

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. OpenAlex reports about 3 citations worldwide. Full citation record

  1. Neuro-Symbolic Injection of LTLf Constraints in Autoregressive Reinforcement Learning Policies

    cs.AI 2026-06 unverdicted novelty 6.0 of 10

    A neuro-symbolic framework compiles LTLf formulas to DFAs, derives differentiable satisfaction signals from DFA progression, and uses them as a logic-based regularization loss to enforce temporal constraints in autore...

Reference graph

Works this paper leans on

50 extracted references · 34 canonical work pages · cited by 1 Pith paper

  1. [1]

    Rama-Maneiro, J

    E. Rama-Maneiro, J. C. Vidal, M. Lama, Deep learning for predictive business process monitoring: Review and benchmark, IEEE Transac- tions on Services Computing 16 (2020) 739–756

    work page 2020

  2. [3]

    Rivera Lazo, R

    G. Rivera Lazo, R. ˜Nanculef, Multi-attribute transformers for sequence prediction in business process management, in: Discovery Science: 25th International Conference, DS 2022, Montpellier, France, October 10–12, 2022, Proceedings, Springer-Verlag, Berlin, Heidelberg, 2022, p. 184–194. doi:10.1007/978-3-031-18840-4_14

    work page doi:10.1007/978-3-031-18840-4_14 2022

  3. [4]

    Rama-Maneiro, F

    E. Rama-Maneiro, F. Patrizi, J. C. Vidal, M. Lama, Towards learning the optimal sampling strategy for suffix prediction in predictive moni- toring, in: Proc. of CAISE 2024, 2024, p. To Appear

    work page 2024

  4. [5]

    Giunchiglia, M

    E. Giunchiglia, M. C. Stoian, S. Khan, F. Cuzzolin, T. Lukasiewicz, ROAD-R: the autonomous driving dataset with logical require- ments, Mach. Learn. 112 (9) (2023) 3261–3291. doi:10.1007/ S10994-023-06322-Z . 22

    work page 2023

  5. [6]

    M. Ma, J. Gao, L. Feng, J. Stankovic, Stlnet: Signal temporal logic enforced multivariate recurrent neural networks, in: H. Larochelle, M. Ranzato, R. Hadsell, M. Balcan, H. Lin (Eds.), Advances in Neural Information Processing Systems, Vol. 33, Curran Associates, Inc., 2020, pp. 14604–14614

    work page 2020

  6. [7]

    Di Francescomarino, F

    C. Di Francescomarino, F. M. Maggi, G. Petrucci, A. Yeshchenko, An eye into the future: Leveraging a-priori knowledge in predictive business process monitoring, in: Business Process Management - 15th Interna- tional Conference, BPM 2017, Barcelona, Spain, September 10-15, 2017, Proceedings, 2017, pp. 252–268. doi:10.1007/978-3-319-65000-5\ _15

    work page doi:10.1007/978-3-319-65000-5 2017

  7. [8]

    Collura, K

    V. Collura, K. Tit, L. Bussi, E. Giunchiglia, M. Cordy, TRIDENT: temporally restricted inference via dfa-enhanced neural traversal, CoRR abs/2506.09701 (2025). arXiv:2506.09701, doi:10.48550/ARXIV. 2506.09701. URL https://doi.org/10.48550/arXiv.2506.09701

    work page doi:10.48550/arxiv.2506.09701 2025

  8. [9]

    X. Lu, P. West, R. Zellers, R. Le Bras, C. Bhagavatula, Y. Choi, Neuro- Logic decoding: (un)supervised neural text generation with predicate logic constraints, in: K. Toutanova, A. Rumshisky, L. Zettlemoyer, D. Hakkani-Tur, I. Beltagy, S. Bethard, R. Cotterell, T. Chakraborty, Y. Zhou (Eds.), Proceedings of the 2021 Conference of the North Amer- ican Chap...

    2021

Show all 50 references
  1. [10]

    X. Lu, S. Welleck, P. West, L. Jiang, J. Kasai, D. Khashabi, R. Le Bras, L. Qin, Y. Yu, R. Zellers, N. A. Smith, Y. Choi, NeuroLogic a*esque decoding: Constrained text generation with lookahead heuristics, in: M. Carpuat, M.-C. de Marneffe, I. V. Meza Ruiz (Eds.), Proceedings ...

    2022

  2. [11]

    Krause, A

    B. Krause, A. D. Gotmare, B. McCann, N. S. Keskar, S. Joty, R. Socher, N. F. Rajani, GeDi: Generative discriminator guided sequence gen- eration, in: M.-F. Moens, X. Huang, L. Specia, S. W.-t. Yih (Eds.), Findings of the Association for Computational Linguistics: EMNLP 2021, A...

    2021 doi

  3. [12]

    Zhang, P.-N

    H. Zhang, P.-N. Kung, M. Yoshida, G. V. den Broeck, N. Peng, Adapt- able logical control for large language models, in: The Thirty-eighth Annual Conference on Neural Information Processing Systems, 2024. URL https://openreview.net/forum?id=58X9v92zRd

    2024

  4. [13]

    N. Miao, H. Zhou, L. Mou, R. Yan, L. Li, Cgmh: constrained sen- tence generation by metropolis-hastings sampling, in: Proceedings of the Thirty-Third AAAI Conference on Artificial Intelligence and Thirty- First Innovative Applications of Artificial Intelligence Conference and ...

    2019 doi

  5. [14]

    Loula, B

    J. Loula, B. LeBrun, L. Du, B. Lipkin, C. Pasti, G. Grand, T. Liu, Y. Emara, M. Freedman, J. Eisner, R. Cotterell, V. Mansinghka, A. K. Lew, T. Vieira, T. J. O’Donnell, Syntactic and semantic control of large language models via sequential monte carlo, in: The Thirteenth Inter...

    2025

  6. [15]

    E. Jang, S. Gu, B. Poole, Categorical reparameterization with gumbel- softmax, in: 5th International Conference on Learning Representations, ICLR 2017, Toulon, France, April 24-26, 2017, Conference Track Pro- ceedings, 2017

    2017

  7. [16]

    F. J. Damerau, A technique for computer detection and correction of spelling errors, Commun. ACM 7 (3) (1964) 171–176. doi:10.1145/ 363958.363994. 24

    1964

  8. [17]

    Umili, G

    E. Umili, G. P. Licks, F. Patrizi, Enhancing deep sequence generation with logical temporal knowledge, in: Proceedings of the 3rd Interna- tional Workshop on Process Management in the AI Era (PMAI 2024) co- located with 27th European Conference on Artificial Intelligence (ECAI...

    2024

  9. [18]

    A. Pnueli, The temporal logic of programs, in: 18th Annual Symposium on Foundations of Computer Science, Providence, Rhode Island, USA, 31 October - 1 November 1977, IEEE Computer Society, 1977, pp. 46–

    1977

  10. [19]

    De Giacomo, M

    G. De Giacomo, M. Y. Vardi, Linear temporal logic and linear dynamic logic on finite traces, in: Proceedings of the Twenty-Third International Joint Conference on Artificial Intelligence, IJCAI ’13, AAAI Press, 2013, p. 854–860

    2013

  11. [20]

    De Giacomo, R

    G. De Giacomo, R. De Masellis, M. Montali, Reasoning on ltl on finite traces: Insensitivity to infiniteness, Proceedings of the AAAI Conference on Artificial Intelligence 28 (1) (Jun. 2014). doi:10.1609/aaai.v28i1. 8872. URL https://ojs.aaai.org/index.php/AAAI/article/view/8872

    2014 doi

  12. [21]

    Touvron, T

    H. Touvron, T. Lavril, G. Izacard, X. Martinet, M. Lachaux, T. Lacroix, B. Rozi` ere, N. Goyal, E. Hambro, F. Azhar, A. Rodriguez, A. Joulin, E. Grave, G. Lample, Llama: Open and efficient foundation language models, CoRR abs/2302.13971 (2023). arXiv:2302.13971, doi:10. 48550/...

    2023 doi arXiv

  13. [22]

    arXiv:2303.08774

    OpenAI, Gpt-4 technical report (2024). arXiv:2303.08774

    2024 arXiv

  14. [23]

    van den Oord, N

    A. van den Oord, N. Kalchbrenner, K. Kavukcuoglu, Pixel recurrent neural networks, in: M. F. Balcan, K. Q. Weinberger (Eds.), Proceedings of The 33rd International Conference on Machine Learning, Vol. 48 of Proceedings of Machine Learning Research, PMLR, New York, New York, US...

    2016

  15. [24]

    Salimans, A

    T. Salimans, A. Karpathy, X. Chen, D. P. Kingma, Pixelcnn++: A pixelcnn implementation with discretized logistic mixture likelihood and other modifications, in: ICLR, 2017

    2017

  16. [25]

    Casolaro, V

    A. Casolaro, V. Capone, G. Iannuzzo, F. Camastra, Deep learning for time series forecasting: Advances and open problems, Information 14 (11) (2023). doi:10.3390/info14110598. URL https://www.mdpi.com/2078-2489/14/11/598

    2023 doi

  17. [26]

    Umili, R

    E. Umili, R. Capobianco, Deepdfa: Automata learning through neu- ral probabilistic relaxations, in: ECAI 2024 - 27th European Confer- ence on Artificial Intelligence, 19-24 October 2024, Santiago de Com- postela, Spain - Including 13th Conference on Prestigious Applica- tions ...

    2024 doi

  18. [27]

    C. J. Maddison, A. Mnih, Y. W. Teh, The concrete distribution: A continuous relaxation of discrete random variables, in: 5th International Conference on Learning Representations, ICLR 2017, Toulon, France, April 24-26, 2017, Conference Track Proceedings, 2017. URL https://open...

    2017

  19. [28]

    Fuggitti, Ltlf2dfa (March 2019)

    F. Fuggitti, Ltlf2dfa (March 2019). doi:10.5281/zenodo.3888410

    2019 doi

  20. [29]

    Umili, R

    E. Umili, R. Capobianco, G. D. Giacomo, Grounding ltlf specifica- tions in image sequences, in: Proceedings of the 20th International Conference on Principles of Knowledge Representation and Reasoning, KR 2023, Rhodes, Greece, September 2-8, 2023, 2023, pp. 668–678. doi:10.249...

    2023 doi

  21. [30]

    Manginas, G

    N. Manginas, G. Paliouras, L. D. Raedt, Nesya: Neurosymbolic au- tomata, CoRR abs/2412.07331 (2024). arXiv:2412.07331, doi:10. 48550/ARXIV.2412.07331. URL https://doi.org/10.48550/arXiv.2412.07331

    2024 doi arXiv

  22. [31]

    Donadello, P

    I. Donadello, P. Felli, C. Innes, F. M. Maggi, M. Montali, Conformance checking of fuzzy logs against declarative temporal specifications, in: Business Process Management - 22nd International Conference, BPM 26 2024, Krakow, Poland, September 1-6, 2024, Proceedings, 2024, pp. 39–

    2024

  23. [32]

    S. Zhu, L. M. Tabajara, J. Li, G. Pu, M. Y. Vardi, Symbolic ltlf synthesis, in: Proceedings of the Twenty-Sixth International Joint Conference on Artificial Intelligence, IJCAI-17, 2017, pp. 1362–1369. doi:10.24963/ijcai.2017/189. URL https://doi.org/10.24963/ijcai.2017/189

    2017 doi

  24. [33]

    Bansal, Y

    S. Bansal, Y. Li, L. M. Tabajara, M. Y. Vardi, Hybrid compositional reasoning for reactive synthesis from finite-horizon specifications, in: The Thirty-Fourth AAAI Conference on Artificial Intelligence, AAAI 2020, The Thirty-Second Innovative Applications of Artificial Intelli...

    2020

  25. [34]

    G. D. Giacomo, M. Favorito, Compositional approach to translate ltlf/ldlf into deterministic finite automata, in: S. Biundo, M. Do, R. Goldman, M. Katz, Q. Yang, H. H. Zhuo (Eds.), Proceedings of the Thirty-First International Conference on Automated Planning and Scheduling, I...

    2021

  26. [35]

    Pesic, W

    M. Pesic, W. M. van der Aalst, A declarative approach for flexible busi- ness processes management, in: Business Process Management Work- shops, 2006

    2006

  27. [36]

    Pesic, H

    M. Pesic, H. Schonenberg, W. M. van der Aalst, Declare: Full sup- port for loosely-structured processes, in: 11th IEEE International En- terprise Distributed Object Computing Conference (EDOC 2007), 2007, pp. 287–287. doi:10.1109/EDOC.2007.14

    2007 doi

  28. [37]

    Westergaard, Better algorithms for analyzing and enacting declar- ative workflow languages using ltl, in: S

    M. Westergaard, Better algorithms for analyzing and enacting declar- ative workflow languages using ltl, in: S. Rinderle-Ma, F. Toumani, 27 K. Wolf (Eds.), Business Process Management, Springer Berlin Heidel- berg, Berlin, Heidelberg, 2011, pp. 83–98

    2011

  29. [38]

    Alman, C

    A. Alman, C. D. Ciccio, D. Haas, F. M. Maggi, A. Nolte, Rule mining with rum, in: ICPM, IEEE, 2020, pp. 121–128

    2020

  30. [39]

    N. Tax, I. Verenich, M. L. Rosa, M. Dumas, Predictive business pro- cess monitoring with LSTM neural networks, in: Advanced Informa- tion Systems Engineering - 29th International Conference, CAiSE 2017, Essen, Germany, June 12-16, 2017, Proceedings, 2017, pp. 477–492. doi:10.1...

    2017 doi

  31. [40]

    Evermann, J

    J. Evermann, J. Rehse, P. Fettke, Predicting process behaviour using deep learning, Decis. Support Syst. 100 (2017) 129–140. doi:10.1016/ J.DSS.2017.04.003. URL https://doi.org/10.1016/j.dss.2017.04.003

    2017 doi

  32. [41]

    Ketyk´ o, F

    I. Ketyk´ o, F. Mannhardt, M. Hassani, B. F. van Dongen, What averages do not tell - predicting real life processes with sequential deep learning, CoRR abs/2110.10225 (2021). arXiv:2110.10225. URL https://arxiv.org/abs/2110.10225

    2021 arXiv

  33. [42]

    N. D. Mauro, A. Appice, T. M. A. Basile, Activity prediction of busi- ness process instances with inception CNN models, in: AI*IA 2019 - Advances in Artificial Intelligence - XVIIIth International Confer- ence of the Italian Association for Artificial Intelligence, Rende, Ital...

    2019 doi

  34. [43]

    Taymouri, M

    F. Taymouri, M. L. Rosa, S. M. Erfani, A deep adversarial model for suffix and remaining time prediction of event sequences, in: Proceedings of the 2021 SIAM International Conference on Data Mining, SDM 2021, Virtual Event, April 29 - May 1, 2021, 2021, pp. 522–530. doi:10. 11...

    2021 doi

  35. [44]

    Chiorrini, C

    A. Chiorrini, C. Diamantini, A. Mircoli, D. Potena, A preliminary study on the application of reinforcement learning for predictive process moni- toring, in: Process Mining Workshops - ICPM 2020 International Work- 28 shops, Padua, Italy, October 5-8, 2020, Revised Selected Pa...

    2020 doi

  36. [45]

    T. R. Besold, A. S. d’Avila Garcez, S. Bader, H. Bowman, P. M. Domingos, P. Hitzler, K. K¨ uhnberger, L. C. Lamb, P. M. V. Lima, L. de Penning, G. Pinkas, H. Poon, G. Zaverucha, Neural-symbolic learning and reasoning: A survey and interpretation, in: Neuro- Symbolic Artificial...

    2021 doi

  37. [46]

    Y. Xie, F. Zhou, H. Soh, Embedding symbolic temporal knowledge into deep sequential models, CoRR abs/2101.11981 (2021). arXiv:2101. 11981. URL https://arxiv.org/abs/2101.11981

    2021 arXiv

  38. [47]

    Badreddine, A

    S. Badreddine, A. d’Avila Garcez, L. Serafini, M. Spranger, Logic tensor networks, Artificial Intelligence 303 (2022) 103649. doi:https://doi.org/10.1016/j.artint.2021.103649. URL https://www.sciencedirect.com/science/article/pii/ S0004370221002009

    2022

  39. [48]

    Umili, F

    E. Umili, F. Argenziano, A. Barbin, R. Capobianco, Visual reward ma- chines, in: Proceedings of the 17th International Workshop on Neural- Symbolic Learning and Reasoning, La Certosa di Pontignano, Siena, Italy, July 3-5, 2023, 2023, pp. 255–267. URL https://ceur-ws.org/Vol-34...

    2023

  40. [49]

    Umili, R

    E. Umili, R. Capobianco, DeepDF A: a transparent neural network design for dfa induction (2023). doi:10.13140/RG.2.2.25449.98401. 29

    2023

  41. [56]

    URL https://doi.org/10.1007/978-3-031-70396-6_3

    doi:10.1007/978-3-031-70396-6\_3 . URL https://doi.org/10.1007/978-3-031-70396-6_3

    doi

  42. [57]

    URL https://doi.org/10.1109/SFCS.1977.32

    doi:10.1109/SFCS.1977.32. URL https://doi.org/10.1109/SFCS.1977.32

    1977 doi

Pith tools

Reviewed August 5, 2026 · model on record in the stance chip above.