mlsanityData to 7 Oct 2026

BEARS Make Neuro-Symbolic Models Aware of their Reasoning Shortcuts

arXiv 2402.12240 ↗First posted 19 Feb 2024cs.LG cs.AIcited by 36

Not on the map. Why?

Abstract (arXiv)

Neuro-Symbolic (NeSy) predictors that conform to symbolic knowledge - encoding, e.g., safety constraints - can be affected by Reasoning Shortcuts (RSs): They learn concepts consistent with the symbolic knowledge by exploiting unintended semantics. RSs compromise reliability and generalization and, as we show in this paper, they are linked to NeSy models being overconfident about the predicted concepts. Unfortunately, the only trustworthy mitigation strategy requires collecting costly dense supervision over the concepts. Rather than attempting to avoid RSs altogether, we propose to ensure NeSy models are aware of the semantic ambiguity of the concepts they learn, thus enabling their users to identify and distrust low-quality concepts. Starting from three simple desiderata, we derive bears (BE Aware of Reasoning Shortcuts), an ensembling technique that calibrates the model's concept-level confidence without compromising prediction accuracy, thus encouraging NeSy architectures to be uncertain about concepts affected by RSs. We show empirically that bears improves RS-awareness of several state-of-the-art NeSy models, and also facilitates acquiring informative dense annotations for mitigation purposes.

Key facts. Cited by 36 papers in the map (data through 7 Oct 2026). Builds on 59 of its 90 references that are in the map. First posted on arXiv 19 Feb 2024.

Builds on

59 of 90 references.

References · 59

  1. 2015Faster R-CNN: Towards Real-Time Object Detection with Region Proposal Networkscited by 9,617
  2. 2015Dropout as a Bayesian Approximation: Representing Model Uncertainty in Deep Learningcited by 4,416
  3. 2016Simple and Scalable Predictive Uncertainty Estimation using Deep Ensemblescited by 3,888
  4. 2017What Uncertainties Do We Need in Bayesian Deep Learning for Computer Vision?cited by 2,335
  5. 2020Concept Bottleneck Modelscited by 1,022
  6. 2019When Does Label Smoothing Help?cited by 985
  7. 2020A Review of Uncertainty Quantification in Deep Learning: Techniques, Applications and Challengescited by 813
  8. 2019Variational Autoencoders and Nonlinear ICA: A Unifying Frameworkcited by 494
  9. 2019Neural Probabilistic Logic Programming in DeepProbLogcited by 389
  10. 2021Laplace Redux -- Effortless Bayesian Deep Learningcited by 359
  11. 2020Calibrating Deep Neural Networks using Focal Losscited by 353
  12. 2020Hands-on Bayesian Neural Networks -- a Tutorial for Deep Learning Userscited by 336
38 more
  1. 2017A Semantic Loss Function for Deep Learning with Symbolic Knowledgecited by 267
  2. 2022Mitigating Neural Network Overconfidence with Logit Normalizationcited by 251
  3. 2020Logic Tensor Networkscited by 197
  4. 2017Neural-Symbolic Learning and Reasoning: A Survey and Interpretationcited by 195
  5. 2019Practical Deep Learning with Bayesian Principlescited by 158
  6. 2019Neural-Symbolic Computing: An Effective Methodology for Principled Integration of Machine Learning and Reasoningcited by 149
  7. 2021Repulsive Deep Ensembles are Bayesiancited by 103
  8. 2022Semantic Probabilistic Layers for Neuro-Symbolic Learningcited by 103
  9. 2023Probabilistic Concept Bottleneck Modelscited by 101
  10. 2020Analyzing Differentiable Fuzzy Logic Operatorscited by 100
  11. 2023Concept-based Explainable Artificial Intelligence: A Surveycited by 97
  12. 2021Independent mechanism analysis, a new concept?cited by 96
  13. 2017Logic Tensor Networks for Semantic Image Interpretationcited by 92
  14. 2023Learning Linear Causal Representations from Interventions under General Nonlinear Mixingcited by 92
  15. 2023NeurASP: Embracing Neural Networks into Answer Set Programmingcited by 92
  16. 2020Explainable Object-induced Action Decision for Autonomous Vehiclescited by 88
  17. 2020On Mixup Regularizationcited by 82
  18. 2022Concept Bottleneck Model with Additional Unsupervised Conceptscited by 79
  19. 2022GlanceNets: Interpretabile, Leak-proof Concept-based Modelscited by 73
  20. 2020Coherent Hierarchical Multi-Label Classification Networkscited by 65
  21. 2023Not All Neuro-Symbolic Concepts Are Created Equal: Analysis and Mitigation of Reasoning Shortcutscited by 63
  22. 2022Concept Embedding Models: Beyond the Accuracy-Explainability Trade-Offcited by 61
  23. 2020Closed Loop Neural-Symbolic Learning via Integrating Neural Perception, Grammar Parsing, and Symbolic Reasoningcited by 58
  24. 2021Incorporating Domain Knowledge into Deep Neural Networkscited by 56
  25. 2023Causal Component Analysiscited by 56
  26. 2022Leveraging Explanations in Interactive Machine Learning: An Overviewcited by 52
  27. 2022A-NeSI: A Scalable Approximate Method for Probabilistic Neurosymbolic Inferencecited by 50
  28. 2021MultiplexNet: Towards Fully Satisfied Logical Constraints in Neural Networkscited by 44
  29. 2022Deep Learning with Logical Constraintscited by 44
  30. 2023Human Uncertainty in Concept-Based AI Systemscited by 39
  31. 2024Nonparametric Partial Disentanglement via Mechanism Sparsity: Sparse Actions, Interventions and Sparse Temporal Dependenciescited by 39
  32. 2020Abductive Knowledge Induction From Raw Datacited by 38
  33. 2023BISCUIT: Causal Representation Learning from Binary Interactionscited by 38
  34. 2021Symbols as a Lingua Franca for Bridging Human-AI Chasm for Explainable and Advisable AI Systemscited by 36
  35. 2022NeuPSL: Neural Probabilistic Soft Logiccited by 34
  36. 2023Neuro-Symbolic Continual Learning: Knowledge, Reasoning Shortcuts and Concept Rehearsalcited by 33
  37. 2022VAEL: Bridging Variational Autoencoders and Probabilistic Logic Programmingcited by 31
  38. 2023A Rigorous Link between Deep Ensembles and (Variational) Bayesian Methodscited by 31

50 most cited of 59 All 59 (JSON)

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2024: 10; 2025: 16; 2026: 10.

36 papers.

Sort:

By area:

  1. 2025Neuro-Symbolic AI in 2024: A Systematic Reviewcited by 36
  2. 2024Logically Consistent Language Models via Neuro-Symbolic Integrationcited by 30
  3. 2024What is the Relationship between Tensor Factorizations and Circuits (and How Can We Exploit it)?cited by 22
  4. 2024Sum of Squares Circuitscited by 21
  5. 2024A Neuro-Symbolic Benchmark Suite for Concept Quality and Reasoning Shortcutscited by 19
  6. 2024On the Independence Assumption in Neurosymbolic Learningcited by 18
  7. 2024Proof of Thought : Neurosymbolic Program Synthesis allows Robust and Interpretable Reasoningcited by 17
  8. 2025Shortcuts and Identifiability in Concept-based Models from a Neuro-Symbolic Lenscited by 13
  9. 2025Neurosymbolic Diffusion Modelscited by 12
  10. 2025A Probabilistic Neuro-symbolic Layer for Algebraic Constraint Satisfactioncited by 11
  11. 2024Neuro-Symbolic AI: Explainability, Challenges, and Future Trendscited by 10
  12. 2026AutoMonitor-Bench: Evaluating the Reliability of LLM-Based Misbehavior Monitorcited by 8
24 more
  1. 2025Neural Probabilistic Circuits: Enabling Compositional and Interpretable Predictions through Logical Reasoningcited by 6
  2. 2025How to Square Tensor Networks and Circuits Without Squaring Themcited by 5
  3. 2024How Do Artificial Intelligences Think? The Three Mathematico-Cognitive Factors of Categorical Segmentation Operated by Synthetic Neuronscited by 3
  4. 2025A Learnability Analysis on Neuro-Symbolic Learningcited by 3
  5. 2025The Road to Generalizable Neuro-Symbolic Learning Should be Paved with Foundation Modelscited by 3
  6. 2025Right for the Right Reasons: Avoiding Reasoning Shortcuts via Prototypical Neurosymbolic AIcited by 3
  7. 2024Imbalances in Neurosymbolic Learning: Characterization and Mitigating Strategiescited by 2
  8. 2024MARS: A neurosymbolic approach for interpretable drug discoverycited by 2
  9. 2025Synthetic Categorical Restructuring large Or How AIs Gradually Extract Efficient Regularities from Their Experience of the Worldcited by 2
  10. 2025A Scalable Approach to Probabilistic Neuro-Symbolic Robustness Verificationcited by 1
  11. 2025Gradient-Based Optimization on G"odel Logic as Discrete Local Searchcited by 1
  12. 2025Curriculum Abductive Learningcited by 1
  13. 2025Intra-neuronal attention within language models Relationships between activation and semanticscited by 0
  14. 2025Neuro-symbolic Weak Supervision: Theory and Semanticscited by 0
  15. 2025Polysemy of Synthetic Neurons Towards a New Type of Explanatory Categorical Vector Spacescited by 0
  16. 2026On Improving Neurosymbolic Learning by Exploiting the Representation Spacecited by 0
  17. 2026Can VLMs Reason Robustly? A Neuro-Symbolic Investigationcited by 0
  18. 2026Constraint-Based Analysis of Reasoning Shortcuts in Neurosymbolic Learningcited by 0
  19. 2026Concepts Worth Having: Refining VLM-Guided Concept Bottleneck Models with Minimal Annotationscited by 0
  20. 2026Concise and Logically Consistent Conformal Sets for Neuro-Symbolic Concept-Based Modelscited by 0
  21. 2026A homotopy-type-theoretic generalization of neurosymbolic inferencecited by 0
  22. 2026Differentiable Logic Programming to Mitigate Reasoning Shortcuts in Neurosymbolic Systemscited by 0
  23. 2026Neuro-symbolic learning over OWL 2 DL via consequence-based compilation to differentiable circuitscited by 0
  24. 2026Soft Symbol Grounding for Prototypical Conceptscited by 0