Pith. sign in

REVIEW 3 major objections 5 minor 33 references

Hybrid mechanistic–data-driven models are, in a precise sense, neuro-symbolic models: mechanism supplies the logic, learning supplies the belief, and every hybrid architecture induces an explicit inference functional.

Reviewed by Pith at T0; open to challenge. T0 means a machine referee read the full paper against a public rubric. the ladder, T0–T4 →

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

T0 review · deepseek-v4-flash

2026-08-01 04:23 UTC pith:J6IMUB4B

load-bearing objection A genuinely useful translation from hybrid modeling to NeSy, with a case study whose headline claim about pre-hoc OOD uncertainty is softer than the abstract admits. the 3 major comments →

arxiv 2607.22811 v1 pith:J6IMUB4B submitted 2026-07-24 cs.LG cs.AIcs.LOstat.ML

From Hybrid Mechanistic--Data-Driven Modeling Toward Neuro-Symbolic AI: What, Why, and How

Moein E. Samadi , Andreas Schuppert This is my paper
classification cs.LG cs.AIcs.LOstat.ML
keywords hybrid modelingneuro-symbolic AImechanistic modelsepistemic uncertaintystructural violation ratebelief dispersionout-of-distribution detectionstructured Boolean networks
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved

The pith

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

Hybrid mechanistic–data-driven models — the standard way to combine first-principles equations with learned components in process engineering and scientific machine learning — are usually described by their architectures and training losses. The paper's central claim is that every such design can be reconstructed, in a principled way, as a neuro-symbolic (NeSy) model: the mechanistic equations and structural constraints define a logical language and its semantics, the learned components define a belief distribution over the unknown quantities, and validity rules and constraints act as logic predicates or as restrictions of the integration domain. This reconstruction, called H2N, yields an explicit inference functional for each design and splits the model into a logic side (what is admissible) and a belief side (what is plausible). From that split the paper derives two metrics — SVR, the belief mass that violates the mechanistic structure, and BD, the dispersion of the learned plausibility — and shows in a structured Boolean-network case study under label noise that these metrics quantify a hybrid model's epistemic uncertainty about its own mechanistic part at deployment time. The payoff is that a model can report how much reliability to expect on uncovered regions during extrapolation, before any out-of-distribution labels are observed, whereas test accuracy notices the same shift only afterwards.

Core claim

The central discovery is a correspondence with measurable consequences: hybrid mechanistic–data-driven models can be reconstructed as neuro-symbolic models in the sense of the NeSy tuple (L, μ, Ω, bθ), where mechanistic equations supply the language L and semantics μ, learned modules induce a belief bθ over unknown functions or parameters, and validity rules and hard constraints are encoded either as a logic function l or as a restriction of the integration domain Ω′. Every hybrid architecture then induces an explicit inference functional of the form Fθ,x(φ) = ∫Ω′ l(φ,ω) bθ,x(ω) dm(ω). Because the admissible/violating partition of Ω′ is set by the logic side alone, two quantities become meas

What carries the argument

The key machinery is the H2N translation together with the logic–belief separation property. H2N maps each hybrid design to a NeSy tuple (L, μ, Ω, bθ) and replaces the usual architecture description with the inference functional Fθ,x(φ)=∫Ω′ l(φ,ω)bθ,x(ω) dm(ω). The load-bearing step is Proposition 3: for a fixed tolerance τ, the partition of the interpretation space into admissible and violating regions is determined entirely by the logic function l — never by the belief bθ. This independence is what lets the paper read BDunseen off the learned row coverage alone, before any out-of-distribution sample arrives, and is what makes SVR and BD decoupled measures (tolerance moves SVR without touch

Load-bearing premise

The deployment-time diagnostic assumes the learned first-layer Boolean modules correctly capture the mechanistic partition: if the learner recovers a label-equivalent but structurally different factorisation, BDunseen can read zero while OOD accuracy is still far below in-distribution accuracy, so the pre-hoc warning is only as good as the learned structure.

What would settle it

Train the structured model on data from a ground-truth partition while deliberately forcing the learner to recover a label-equivalent but different first-layer partition; if BDunseen stays near zero while out-of-distribution accuracy falls far below in-distribution accuracy across many seeds, the claim that BDunseen quantifies extrapolation uncertainty before OOD labels are observed is falsified.

Watch this falsifier — get emailed when new claim-graph text bears on it.

If this is right

  • Any two hybrid designs become comparable as explicit NeSy objects: the (L, μ, Ω, bθ) tuple and the induced inference functional replace architecture-and-loss descriptions as the shared semantic interface.
  • SVR and BD can be computed at deployment time from training data alone, so a trained model can report the reliability of its mechanistic assumptions before any labeled test or out-of-distribution data arrive.
  • Because BDunseen is computed from the coverage of the integration domain by the learned partition, extrapolation risk on uncovered regimes is signalled in advance rather than after accuracy collapses.
  • Under a fixed belief, changing the noise-tolerance budget κ moves SVR across its full range while leaving BD and test accuracy unchanged, making noise tolerance a design choice of the logic side.
  • In the case study, seed-level variability of held-out accuracy rises monotonically with BD, so BD is an a priori proxy for how unpredictable a trained model's accuracy will be.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The same translation should extend to continuous hybrid models (neural ODEs with mechanistic balance equations), where BDunseen would quantify coverage of the state space actually reached; this is a testable extension beyond the paper's Boolean case study.
  • The SVR/BD decoupling suggests a decision rule the paper leaves implicit: underperformance with high SVR points to a faulty mechanistic structure, while high BD points to missing data or identifiability problems — the two metrics split diagnosis from remedy.
  • BDunseen could be used as a model-selection criterion among candidate learned structures; the paper's own caveat that a zero BDunseen does not certify OOD robustness implies such a selection rule would need to be paired with a check that the learned partition respects the true mechanistic groupings.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

3 major / 5 minor

Summary. The paper proposes Hybrid-to-NeSy (H2N), a translation procedure that reconstructs hybrid mechanistic/data-driven models as neuro-symbolic models in the sense of De Smet and De Raedt. It places mechanistic equations and structural constraints on the logic side, learned components on the belief side, and expresses validity domains and constraints through logic functions and restricted integration domains. From the resulting logic–belief decomposition, the paper derives two metrics: SVR (structural violation rate) and BD (belief dispersion). The method is instantiated on a structured Boolean classification model with label noise, where SVR and BD are reported to track variability in held-out accuracy, and an OOD experiment claims that BD_unseen provides a deployment-time indicator of uncertainty during extrapolation.

Significance. If accepted as a conceptual framework, H2N is a useful bridge between hybrid process modeling and neuro-symbolic AI: it gives an explicit inference functional (Eq. 1), a mapping table for canonical hybrid patterns (Table 1), and a clear separation of admissibility from plausibility. The case study is transparent and reproducible, with a closed-form BD, Monte Carlo stability checks, and an honest appendix discussing seed variability. However, the empirical claims currently outrun the evidence: the pre-hoc OOD diagnostic is conditional on the learned structural partition being faithful, and the metrics are not calibrated or compared with established uncertainty-quantification baselines. The central translation idea is defensible, but the paper's headline claims need to be either substantially qualified or supported by additional analysis.

major comments (3)
  1. [Abstract; §5, Result 2; Appendix F] The claim that H2N "quantifies a model's uncertainty during extrapolations" is stronger than the evidence. BD_unseen is computed from row coverage under the learned first-layer factorization, which enters the belief as Dirac factors. Appendix F concedes that in 3 of 10 seeds at N_ood = 4 and 6 the learner recovers a label-equivalent factorization that maps held-out ground-truth rows into observed learned classes, giving BD_unseen = 0. Pooling seeds, BD_unseen = 0 models average OOD accuracy 0.69 versus 0.39 for BD_unseen > 0, while in-distribution accuracy is 0.998. Thus a zero BD_unseen does not indicate safety, and the diagnostic quantifies the model's self-reported coverage under the learned structure, not the actual epistemic state about the OOD region. The abstract and Result 2 should be rephrased as a conditional statement, or the framework should place uncertainty over the first-l
  2. [§5, Result 1; Table 2; Eq. (7)] BD is validated only by a Spearman correlation (rho = -0.94) and a monotone seed-level standard-deviation plot. This is a correlational claim, not a calibration analysis. No evidence is provided that a nominal credible interval constructed from the row-Bernoulli belief has correct frequentist coverage, and no comparison is made with standard UQ baselines (e.g., ensembles, MC dropout, Bayesian neural networks) on the same task. The statement that BD provides "an a priori estimate of how uncertain a trained model's accuracy will be" therefore needs either quantitative calibration diagnostics or a more modest wording.
  3. [Proposition 3; Remark 4; §5] The decomposition BD = BD_seen + BD_unseen and the logic–belief separation are true by construction: with Dirac first-layer modules and independent row-Bernoulli factors, BD = sum_K p_K(1 - p_K) and BD_unseen = (1/4)|{K: n_0(K)+n_1(K)=0}|. This is a definitional identity from the factorized belief, not an independent empirical finding. The paper should clearly distinguish the definitional content of Prop. 3 from empirical validation. As written, the "measurable consequence" framing invites the circularity objection that the metric reproduces its own definition; the conceptual contribution is the placement of the row partition on the logic side, which should be stated as a modeling choice rather than a discovered result.
minor comments (5)
  1. [§5, before Table 2] The notation "struct [4,4,4,4]" is used without definition in the main text; define it formally when the case study is introduced.
  2. [§5, Eq. (belief factorization)] The maximum-entropy prior p_K = 1/2 for unsupported second-layer rows is an assumption. The sensitivity of BD_unseen to this prior should be discussed explicitly; with a different prior, the decomposition and the numeric values change.
  3. [Appendix G, Figure G1] The normalization of SVR and BD to [0,1] in Figure G1 is not specified. State the exact normalization used so the reader can interpret the overlays.
  4. [Title and terminology] The phrase "structural distribution shift" is used for holding out reachable second-layer rows. This is a combinatorial coverage shift rather than a conventional distribution shift; consider using "coverage shift" or "structural extrapolation" to avoid ambiguity.
  5. [Table 1] The header "L µ/l" is awkward; reformat the columns so that the language and logic-function entries are clearly separated.

Circularity Check

0 steps flagged

No significant circularity: the H2N translation is a constructive definition, and the case-study metrics are tested against independent synthetic ground truth; apparent tautologies are explicit definitions, not disguised predictions.

full rationale

The central H2N claim is a representation theorem: given a hybrid model, the paper constructs a NeSy tuple and an inference functional that reproduces the model by design. This is a definitional mapping, not a prediction, so it is not circular. Proposition 3 and the BD = BDseen + BDunseen decomposition are algebraic identities that follow from the definitions of the logic function and the factorized Bernoulli belief; the paper presents them as such, not as empirical discoveries. The empirical content is the case study: BD and SVR are computed from training-label counts and then correlated with held-out accuracy variability and OOD accuracy. These are independent measurements on synthetic data with known ground truth, and the correlation can fail (Appendix F reports seeds where BDunseen = 0 while OOD accuracy is low). Self-citations (NoiseCut, structured hybrid models) are used as tools and background, not as load-bearing uniqueness theorems, and NoiseCut is a code-reproduced package. The Appendix F limitation is a validity caveat about conditioning on the learned partition, not a circularity. Hence no circular step meets the evidentiary bar.

Axiom & Free-Parameter Ledger

2 free parameters · 6 axioms · 0 invented entities

The formal core rests on the borrowed NeSy frame and on case-study-specific modeling choices: Dirac first-layer modules, maximum-entropy unsupported rows, and hand-set κ. No new physical entities are introduced; SVR and BD are derived metrics, not fitted constants.

free parameters (2)
  • Violation budget κ (tolerance τ) = κ=⌊0.5S⌋ (noise sweep); κ=⌊0.10S⌋ (OOD); κ/S swept in [0,0.5]
    SVR is defined relative to κ and changes across its full range as κ moves, while BD is unaffected. The choice of κ is made by the experimenter, not estimated from data.
  • Maximum-entropy prior for unsupported second-layer rows = p_K = 0.5
    Rows with no training support are assigned p_K=1/2 by assumption. This drives BDunseen and is a modeling choice, not a measured quantity; other priors would give different BDunseen values.
axioms (6)
  • domain assumption The De Smet & De Raedt NeSy tuple (L, μ, Ω, b) and integral Eq. (1) can serve as a universal semantic frame for hybrid mechanistic/data-driven models.
    The H2N translation is defined relative to this frame; no proof is given that every hybrid model admits such a representation without loss. Invoked throughout Sections 2–3.
  • standard math Measurability and integrability conditions in Appendix A hold for l(φ,·) and B_{θ,x}.
    Needed for the inference functional to be well-defined; standard but not verified in the case study.
  • domain assumption The hybrid model has a fixed solver when G(x,z;α,ψ)=0 has multiple solutions.
    Section 2, Definition 1: the predictive map assumes a fixed solver, excluding solution-selection ambiguity.
  • domain assumption Learning the first-layer modules via NoiseCut gives the ground-truth Boolean functions (Dirac factors).
    Section 5 places Dirac belief on F_m; Appendix F documents label-equivalent factorizations where this fails, so the assumption is not guaranteed.
  • ad hoc to paper The row partition is logical: uncovered rows are exactly those with zero training support under the learned first-layer functions, so BD decomposes additively and BDunseen is readable at deployment.
    Used for Result 2 (Section 5, Appendix F). It holds relative to the learned factorization, not necessarily the ground truth.
  • ad hoc to paper Unobserved second-layer rows get maximum-entropy prior p_K=1/2.
    Assigned in Section 5 for rows with n0+n1=0; this choice fixes BDunseen and is not independently justified beyond maximum entropy.

pith-pipeline@v1.3.0-alltime-deepseek · 14827 in / 15758 out tokens · 149842 ms · 2026-08-01T04:23:39.828010+00:00 · methodology

0 comments
read the original abstract

Hybrid mechanistic/data-driven models, which combine first-principles with learned components, are increasingly used in process engineering and scientific machine learning. Common hybrid modeling designs are specified primarily through their architectures and training losses, which offers a limited basis for a shared semantic interface to compare or verify them across domains, with comparatively little attention paid to epistemic uncertainty in the mechanistic part. We bridge hybrid modeling and neuro-symbolic (NeSy) AI by reconstructing these designs as instances of NeSy interface. The resulting translation, Hybrid-to-NeSy (H2N), places mechanistic knowledge on the language side, learned modules on the belief side, and validity domains together with constraints on the logic side. For each design, H2N then yields an explicit NeSy inference functional and a logic-belief decomposition. From this decomposition we derive two metrics: structural violation rate (SVR), measuring whether the learned belief respects the mechanistic structure; and belief dispersion (BD), measuring how concentrated the learned plausibility is, serving as a hybrid model's epistemic uncertainty in its mechanistic part. We instantiate H2N on a case study of a structured hybrid model for binary classification under label noise and show that models with higher SVR and BD exhibit greater variability in held-out accuracy. Under structural distribution shift, H2N further quantifies a model's uncertainty during extrapolations, whereas test accuracy reveals the same shift only post hoc.

Figures

Figures reproduced from arXiv: 2607.22811 by Andreas Schuppert, Moein E. Samadi.

Figure 1
Figure 1. Figure 1: Noise sweep, 50 seeds per level (struct [4 [PITH_FULL_IMAGE:figures/full_fig_p008_1.png] view at source ↗

discussion (0)

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

Reference graph

Works this paper leans on

33 extracted references · 3 linked inside Pith

  1. [1]

    Hybrid semi-parametric mathematical systems: Bridging the gap between systems biology and process engineering.Journal of biotechnology, 132(4):418–425, 2007

    Ana P Teixeira, Nuno Carinhas, Jo˜ ao ML Dias, Pedro Cruz, Paula M Alves, Manuel JT Carrondo, and Rui Oliveira. Hybrid semi-parametric mathematical systems: Bridging the gap between systems biology and process engineering.Journal of biotechnology, 132(4):418–425, 2007

    2007

  2. [2]

    Grey-box modelling and identification using physical knowledge and bayesian techniques.Automatica, 29(2):285–308, 1993

    Herbert JAF Tulleken. Grey-box modelling and identification using physical knowledge and bayesian techniques.Automatica, 29(2):285–308, 1993

    1993

  3. [3]

    A hybrid neural network-first principles approach to process modeling.AIChE Journal, 38(10):1499–1511, 1992

    Dimitris C Psichogios and Lyle H Ungar. A hybrid neural network-first principles approach to process modeling.AIChE Journal, 38(10):1499–1511, 1992

    1992

  4. [4]

    Modeling chemical processes using prior knowledge and neural networks.AIChE Journal, 40(8):1328–1340, 1994

    Michael L Thompson and Mark A Kramer. Modeling chemical processes using prior knowledge and neural networks.AIChE Journal, 40(8):1328–1340, 1994

    1994

  5. [5]

    Strategy for dynamic process modeling based on neural networks in macroscopic balances

    Henricus JL Van Can, Chris Hellinga, Karel Ch AM Luyben, Joseph J Heijnen, and Hubert AB Te Braake. Strategy for dynamic process modeling based on neural networks in macroscopic balances. AIChE Journal, 42(12):3403–3418, 1996

    1996

  6. [6]

    Combining neural and conventional paradigms for modelling, prediction and control

    Mukul Agarwal. Combining neural and conventional paradigms for modelling, prediction and control. International Journal of Systems Science, 28(1):65–81, 1997

    1997

  7. [7]

    Extrapolability of structured hybrid models: a key to optimization of complex processes

    Andreas A Schuppert. Extrapolability of structured hybrid models: a key to optimization of complex processes. InEquadiff 99: (In 2 Volumes), pages 1135–1151. World Scientific, 2000

    2000

  8. [8]

    Elsevier, 2001

    Ian T Cameron and Katalin Hangos.Process modelling and model analysis, volume 4. Elsevier, 2001

    2001

  9. [9]

    Local identification of scalar hybrid models with tree structure

    Bernold Fiedler and Andreas Schuppert. Local identification of scalar hybrid models with tree structure. IMA Journal of Applied Mathematics, 73(3):449–476, 2008

    2008

  10. [10]

    Understanding and applying the extrapolation properties of serial gray-box models.AIChE journal, 44(5):1071–1089, 1998

    Henricus JL Van Can, Hubert AB Te Braake, Sander Dubbelman, Chris Hellinga, Karel Ch AM Luyben, and Joseph J Heijnen. Understanding and applying the extrapolation properties of serial gray-box models.AIChE journal, 44(5):1071–1089, 1998

    1998

  11. [11]

    Hybrid semi-parametric modeling in process systems engineering: Past, present and future.Computers & Chemical Engineering, 60:86–101, 2014

    Moritz Von Stosch, Rui Oliveira, Joana Peres, and Sebasti˜ ao Feyo De Azevedo. Hybrid semi-parametric modeling in process systems engineering: Past, present and future.Computers & Chemical Engineering, 60:86–101, 2014

    2014

  12. [12]

    Hybrid neural network modeling of a full-scale industrial wastewater treatment process.Biotechnology and bioengineering, 78(6):670–682, 2002

    Dae Sung Lee, Che Ok Jeon, Jong Moon Park, and Kun Soo Chang. Hybrid neural network modeling of a full-scale industrial wastewater treatment process.Biotechnology and bioengineering, 78(6):670–682, 2002. 9

    2002

  13. [13]

    First-principles, data-based, and hybrid modeling and optimiza- tion of an industrial hydrocracking unit.Industrial & engineering chemistry research, 45(23):7807–7816, 2006

    N Bhutani, GP Rangaiah, and AK Ray. First-principles, data-based, and hybrid modeling and optimiza- tion of an industrial hydrocracking unit.Industrial & engineering chemistry research, 45(23):7807–7816, 2006

    2006

  14. [14]

    Knowledge based modular networks for process modelling and control.Computers & Chemical Engineering, 25(4-6):783–791, 2001

    J Peres, R Oliveira, and S Feyo De Azevedo. Knowledge based modular networks for process modelling and control.Computers & Chemical Engineering, 25(4-6):783–791, 2001

    2001

  15. [15]

    Bioprocess hybrid parametric/nonparametric modelling based on the concept of mixture of experts.Biochemical Engineering Journal, 39(1):190–206, 2008

    J Peres, R Oliveira, and S Feyo de Azevedo. Bioprocess hybrid parametric/nonparametric modelling based on the concept of mixture of experts.Biochemical Engineering Journal, 39(1):190–206, 2008

    2008

  16. [16]

    Fuzzy identification of systems and its applications to modeling and control.IEEE transactions on systems, man, and cybernetics, (1):116–132, 1985

    Tomohiro Takagi and Michio Sugeno. Fuzzy identification of systems and its applications to modeling and control.IEEE transactions on systems, man, and cybernetics, (1):116–132, 1985

    1985

  17. [17]

    A structured modeling approach for dynamic hybrid fuzzy-first principles models.Journal of Process Control, 12(5):605–615, 2002

    Pascal F van Lith, Ben HL Betlem, and Brian Roffel. A structured modeling approach for dynamic hybrid fuzzy-first principles models.Journal of Process Control, 12(5):605–615, 2002

    2002

  18. [18]

    The validity domain of hybrid models and its application in process optimization.Chemical Engineering and Processing: Process Intensification, 46(11):1054–1066, 2007

    Olaf Kahrs and Wolfgang Marquardt. The validity domain of hybrid models and its application in process optimization.Chemical Engineering and Processing: Process Intensification, 46(11):1054–1066, 2007

    2007

  19. [19]

    Efficient reengineering of meso-scale topologies for functional networks in biomed- ical applications.Journal of Mathematics in Industry, 1(1):6, 2011

    Andreas A Schuppert. Efficient reengineering of meso-scale topologies for functional networks in biomed- ical applications.Journal of Mathematics in Industry, 1(1):6, 2011

    2011

  20. [20]

    A hybrid modeling framework for gen- eralizable and interpretable predictions of icu mortality across multiple hospitals.Scientific reports, 14(1):5725, 2024

    Moein E Samadi, Jorge Guzman-Maldonado, Kateryna Nikulina, Hedieh Mirzaieazar, Konstantin Shara- futdinov, Sebastian Johannes Fritsch, and Andreas Schuppert. A hybrid modeling framework for gen- eralizable and interpretable predictions of icu mortality across multiple hospitals.Scientific reports, 14(1):5725, 2024

    2024

  21. [21]

    CRC Press, 2018

    Jarka Glassey and Moritz Von Stosch.Hybrid modeling in process industries. CRC Press, 2018

    2018

  22. [22]

    Dimensions of neural-symbolic integration-a structured survey

    Sebastian Bader and Pascal Hitzler. Dimensions of neural-symbolic integration-a structured survey. arXiv preprint cs/0511042, 2005

    Pith/arXiv arXiv 2005

  23. [23]

    Neural-symbolic learning and reasoning: A survey and interpre- tation.Frontiers in artificial intelligence and applications, 342:1–51, 2022

    P Hitzler, MK Sarker, TR Besold, AD Garcez, S Bader, H Bowman, P Domingos, P Hitzler, KU K¨ uhnberger, LC Lamb, et al. Neural-symbolic learning and reasoning: A survey and interpre- tation.Frontiers in artificial intelligence and applications, 342:1–51, 2022

    2022

  24. [24]

    From statistical relational to neurosymbolic artificial intelligence: A survey.Artificial Intelligence, 328:104062, 2024

    Giuseppe Marra, Sebastijan Dumanˇ ci´ c, Robin Manhaeve, and Luc De Raedt. From statistical relational to neurosymbolic artificial intelligence: A survey.Artificial Intelligence, 328:104062, 2024

    2024

  25. [25]

    Neural-symbolic computing: An effective methodology for principled integration of machine learning and reasoning.arXiv preprint arXiv:1905.06088, 2019

    Artur d’Avila Garcez, Marco Gori, Luis C Lamb, Luciano Serafini, Michael Spranger, and Son N Tran. Neural-symbolic computing: An effective methodology for principled integration of machine learning and reasoning.arXiv preprint arXiv:1905.06088, 2019

    Pith/arXiv arXiv 1905

  26. [26]

    Neurosymbolic ai: The 3 rd wave.Artificial Intelligence Review, 56(11):12387–12406, 2023

    Artur d’Avila Garcez and Luis C Lamb. Neurosymbolic ai: The 3 rd wave.Artificial Intelligence Review, 56(11):12387–12406, 2023

    2023

  27. [27]

    Modular design patterns for hybrid learning and reasoning systems: a taxonomy, patterns and use cases

    Michael van Bekkum, Maaike de Boer, Frank van Harmelen, Andr´ e Meyer-Vitali, and Annette ten Teije. Modular design patterns for hybrid learning and reasoning systems: a taxonomy, patterns and use cases. Applied Intelligence, 51(9):6528–6546, 2021

    2021

  28. [28]

    Probabilistic inference in hybrid domains by weighted model integration

    Vaishak Belle, Andrea Passerini, and Guy Van den Broeck. Probabilistic inference in hybrid domains by weighted model integration. InProceedings of the Twenty-Fourth International Joint Conference on Artificial Intelligence, IJCAI 2015, Buenos Aires, Argentina, July 25-31, 2015, pages 2770–2776. IJCAI Inc, 2015

    2015

  29. [29]

    Not all neuro-symbolic con- cepts are created equal: Analysis and mitigation of reasoning shortcuts.Advances in Neural Information Processing Systems, 36:72507–72539, 2023

    Emanuele Marconato, Stefano Teso, Antonio Vergari, and Andrea Passerini. Not all neuro-symbolic con- cepts are created equal: Analysis and mitigation of reasoning shortcuts.Advances in Neural Information Processing Systems, 36:72507–72539, 2023. 10

    2023

  30. [30]

    Defining neurosymbolic ai.arXiv preprint arXiv:2507.11127, 2025

    Lennert De Smet and Luc De Raedt. Defining neurosymbolic ai.arXiv preprint arXiv:2507.11127, 2025

    Pith/arXiv arXiv 2025

  31. [31]

    Combining prior knowledge with data driven mod- eling of a batch distillation column including start-up.Computers & chemical engineering, 27(7):1021– 1030, 2003

    Pascal F van Lith, Ben HL Betlem, and Brian Roffel. Combining prior knowledge with data driven mod- eling of a batch distillation column including start-up.Computers & chemical engineering, 27(7):1021– 1030, 2003

    2003

  32. [32]

    A training strategy for hybrid models to break the curse of dimensionality.Plos one, 17(9):e0274569, 2022

    Moein E Samadi, Sandra Kiefer, Sebastian Johaness Fritsch, Johannes Bickenbach, and Andreas Schuppert. A training strategy for hybrid models to break the curse of dimensionality.Plos one, 17(9):e0274569, 2022

    2022

  33. [33]

    unseen frac

    Moein E Samadi, Hedieh Mirzaieazar, Alexander Mitsos, and Andreas Schuppert. Noisecut: a python package for noise-tolerant classification of binary data using prior knowledge integration and max-cut solutions.BMC bioinformatics, 25(1):155, 2024. A Measurability assumptions for the inference functional The functional (1) is well-defined under the following...

    2024