Pith. sign in

REVIEW 4 major objections 5 minor 32 references

Think before You Simulate: Symbolic Reasoning to Orchestrate Neural Computation for Counterfactual Question Answering

T0 review · 4 major / 5 minor · reviewed 2026-08-07 · deepseek-v4-flash

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

Pith's one-line read The paper argues that symbolic reasoning about a causal graph should decide when a video-question-answering system simulates and when it trusts perception, and shows this improves counterfactual answers on CLEVRER and CRAFT.

desk verdict A genuinely new neuro-symbolic idea—using symbolic causality to decide when perception is safe—with real gains, but the approximate version's central shortcut has an acknowledged, unquantified failure mode. read the letter →

arxiv 2506.10753 v1 pith:OW5MCLMX submitted 2025-06-12 cs.AI

Adam Ishay , Zhun Yang , Joohyung Lee , Ilgu Kang , Dongjae Lim This is my paper · ORCID
classification cs.AI
keywords counterfactualquestionansweringneuro-symbolicsystemscausalgraphAnswerSetProgrammingvideodynamicsCLEVRERCRAFTperception-simulationorchestration
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 establish that a neuro-symbolic video-question-answering system can be made substantially better at counterfactual questions by deciding, symbolically, when to trust what it saw in the video and when to run a physics simulation. Its central claim is that after an object is removed, an object's simulated trajectory should be replaced by its observed trajectory until the moment the removal could actually affect it, and that this moment can be read off a causal graph of collisions and temporal links. The paper implements the graph computation in Answer Set Programming and combines it with existing perception and simulation modules, reporting state-of-the-art counterfactual accuracy on CLEVRER and improved results on CRAFT when a large language model serves as the simulator. The method is designed to help exactly when perception is more reliable than simulation, which the paper argues is the typical case, and it is modular enough that better perception or better simulation can be plugged in for further gains.

What carries the argument

The central object is the causal graph, a directed graph whose nodes are object states at collision frames, with horizontal edges connecting two objects that collide at a frame and vertical edges connecting one object's state across consecutive collision frames. From this graph Answer Set Programming computes three derived relations: ancestor, affected, and sim node. A node is affected if an intervened object is one of its ancestors; the sim node for an unremoved object is its earliest affected node, which is exactly the frame where the enhanced simulator must switch from perception states to simulated states. In the black-box approximation, the same graph yields a determined verdict: a counterfactual question is determined yes when the queried collision already happened and both objects were unaffected at that frame, and determined no when a queried object is removed or no such collision happened and both were unaffected. The machinery's job is to avoid blind simulation from the first frame, trusting perception until the causal graph says the intervention could have reached the object.

What would settle it

The paper even supplies the falsifying scenario in Figure 5: remove a blue object and ask whether purple and cyan collide; the graph says the collision is unaffected and determines yes, but if the removed blue object would have collided with green, deflecting green into cyan, the true counterfactual has no purple-cyan collision. A reader could go look for such a case in a physical simulator or real video: whenever the determined verdict disagrees with a ground-truth simulation that models multi-step deflections, the over-approximation is exposed. A systematic version would build a small set of CLEVRER-like videos where the removed object's influence reaches the queried objects only through an intermediate object, and then compare CRCGapprox's determined answers to the actual physics.

Watch

Extended reading notes

Core claim

In the paper's own terms, the discovery is that symbolic reasoning about causal structure should orchestrate the use of neural perception and neural simulation, rather than merely aggregating their outputs at the end. The authors define a causal graph whose nodes are object states at the frames where collisions occur, with horizontal edges for collisions and vertical edges for the same object across time. A node is affected by the removed objects if an intervened object is one of its ancestors in this graph; the earliest affected node for each remaining object is its simulation node, and the enhanced simulator uses observed states up to that node and simulated states afterwards. For settings without a frame-by-frame simulator, the paper defines a counterfactual question as determined when the answer can be read directly from unaffected perception states, and it overrides the baseline prediction in those cases. This leads to the reported state-of-the-art counterfactual performance on CLEVRER and to gains on CRAFT, where GPT-3.5 and GPT-4 serve as proxy simulators and the causal graph decides whether a counterfactual question can be safely rephrased as a factual one.

Load-bearing premise

The load-bearing premise is that the causal graph's affected relation, computed as ancestor reachability through collision and temporal edges, marks exactly which object states change under the intervention; the paper itself notes this is an over-approximation that can be wrong even with perfect perception, giving a case where a chain reaction through a third object would change the answer.

Editorial extensions

If this is right

  • If the central claim holds, any neuro-symbolic video-QA system gains from CRCG whenever its perception module is more accurate than its simulator, which the paper says is the usual situation.
  • On CLEVRER counterfactual questions, CRCG with VRDP's perception and simulation modules reaches higher option and question accuracy than the VRDP baseline and the reported prior models.
  • Applying the approximation to NS-DR lifts accuracy on the roughly twenty thousand question-option pairs whose answers are determined by perception, while leaving the not-determined cases to the baseline.
  • On CRAFT, using a GPT-x language model as the simulator, CRCG improves counterfactual question accuracy, and on the subset whose answers are determined it reaches 97.96% and 99.00% accuracy on the easy and hard splits.
  • Because the architecture is modular, the same causal-graph layer can be combined with improved perception and improved simulation modules, which the paper demonstrates by adding its IOD and SPS modules for the other CLEVRER question types.

Reading between the lines

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

  • A testable consequence the paper does not run: vary the quality of the perception and simulation modules across datasets and check whether the size of the CRCG gain tracks the perception-minus-simulation accuracy gap; the paper's own success condition predicts it should.
  • The determined predicate suggests a general recipe for LLM prompting: when a causal model shows a counterfactual question is insensitive to the intervention, re-ask it as a factual question; this could transfer beyond CRAFT to other LLM tasks with an external world model.
  • The paper's Figure 5 counterexample points to a stress test: construct videos where the removed object, though absent, would have caused a chain reaction through a third object before reaching the queried pair; CRCGapprox would answer from perception while true physics says otherwise, so a benchmark of such cases would quantify how often the over-approximation bites.
  • The causal-graph computation is cheap enough that the orchestration layer could plausibly run at inference time to decide whether to trust a simulator, a perception module, or an LLM in a larger video-reasoning system.
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

4 major / 5 minor

Summary. The paper proposes CRCG, a neuro-symbolic architecture for counterfactual question answering that uses Answer Set Programming (ASP) to build a causal graph from perceived collisions and temporal object states, then uses this graph to decide when to trust perception states over simulation states. For cases where only a black-box simulator (or none) is available, it introduces CRCG_approx, which marks a counterfactual query as 'determined' when the answer can be read directly from the perceived video under the causal graph, overriding the baseline prediction. Experiments on CLEVRER report state-of-the-art counterfactual question accuracy (CRCG_VRDP 87.8% vs. VRDP 84.3%) and on CRAFT show improved GPT-4 accuracy (from 77.93% to 79.68% on the easy split). The implementation is publicly available.

Significance. If the results hold, the core idea is valuable: using symbolic causal reasoning to orchestrate when to use perception versus simulation is a modular, interpretable enhancement that can be applied to existing neuro-symbolic systems and even to LLM-based baselines. The paper ships a public implementation and reports gains on two benchmarks, with ablations showing where the gains come from. The main caveat is that CRCG_approx's 'determined' criterion is acknowledged to be an over-approximation that can fail even with perfect perception; the empirical support would be stronger if this failure mode were quantified and separated from perception noise.

major comments (4)
  1. [Section 3.3, Definition 4 and Remark; Figure 5] The load-bearing step for all CRCG_approx results (Tables 2, 4, 5) is the 'determined' criterion, yet the paper's own Remark admits that this criterion can be wrong even with perfect perception: a cascade collision invisible in the original video can change a queried object's state, as in Figure 5. The paper states such cases 'don't occur often in practice' but provides no quantitative estimate. Since Table 2 shows that determined predictions are only 95.08% accurate, the 4.9% error rate needs a breakdown: how many errors come from this cascade over-approximation versus from perception noise? Without this breakdown, the improvement attributed to causal-graph reasoning is not isolated from the baseline simulation errors it is meant to correct.
  2. [Section 5.1, Table 2 and Section 6, Tables 4-5] The accuracy gains for determined options are reported without error bars, confidence intervals, or significance tests. Given that the collision threshold (23.0) and temporal resolution are tuned on the validation set, and the CRAFT GPT experiments use a single temperature-0 run, the reader cannot assess whether the improvements are stable. The paper should report variance across multiple runs or bootstrap resampling, particularly for the 20,558-option CLEVRER subset and the 1,128-question CRAFT subset.
  3. [Section 6 and Appendix C] The text says GPT-x is used as a 'proxy for a simulator' (Ms in Figure 4), but in CRCG_approx the simulator is never invoked for determined cases; GPT-x is only used as the baseline predictor for undetermined cases. This is a conceptual mismatch that should be clarified. The CRCG-guided-prompt variant is a different mechanism (rephrasing the query as a perception question) and should be described as such rather than as simulation.
  4. [Section 5.2, Table 3 and Table 9] The state-of-the-art claim on CLEVRER is supported by CRCG together with additional IOD and SPS modules, not by CRCG alone. The ablation in Table 9 is helpful, but the paper should state more explicitly which components are responsible for the headline SOTA number and should report whether the SOTA gap over VRDP is statistically meaningful, especially since the gains on non-counterfactual question types come largely from the simple post-processing modules rather than from causal reasoning.
minor comments (5)
  1. [Abstract and Section 5.2] The abstract attributes 'state-of-the-art performance' to the enhancement without mentioning that the CLEVRER SOTA result uses the additional IOD and SPS modules; please qualify the claim.
  2. [Section 3.3, Definition 4] The prose for 'determined to be no' leaves the quantifier over t implicit: 'st_o1 and st_o2 are not affected' should read 'for all frames t, neither object's state is affected at t' (or equivalent), as the ASP code in Appendix B makes clear.
  3. [Section 6 versus Appendix C] The model name and version for GPT-x are inconsistent: Section 6 says 'gpt-3.5-turbo-0613' and 'gpt-4-0613' with temperature 0, while Appendix C says 'text-davinci-002'. Please reconcile.
  4. [Table 3] The column headers in Table 3 are misaligned; there should be separate 'opt.' and 'ques.' columns for each question type, and the current rendering makes several numbers ambiguous.
  5. [Section 3.3, footnote 3] The footnote says 'See Table 2 for illustration,' but Table 2 does not directly illustrate the rarity of cascade failures; it only reports determined-option accuracy. A direct example or a small case count would be more helpful.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: CRCG is a decision-orchestration algorithm; the determined criterion is an openly acknowledged approximation, not a tautology.

full rationale

The paper's central mechanism is a causal graph built from perceived in-video collisions and intervened objects, with ASP computing affected and sim nodes. The 'determined' answer is a stipulated decision rule—use the perceived collision when both queried states are graph-unaffected—not a claim that this rule is derived from first principles. The authors explicitly flag the rule's over-approximation in the Remark of Section 3.3 and Figure 5, showing it can fail even with perfect perception; this is a correctness limitation, not a definitional equivalence. No parameter is fitted to the counterfactual target: the only learned constant, the SPS collision threshold (23.0, learned from validation), belongs to an auxiliary physics simulator and is not part of the causal-graph computation, which is parameter-free. The NeurASP self-citation, if included, appears only as related work and is not load-bearing. The method is benchmarked on external datasets (CLEVRER, CRAFT) against independent baselines, so its gains are not forced by construction.

Assumptions & free parameters 2 free parameters · 4 assumptions · 0 invented entities

The central method has no free parameters; the only fitted numeric quantities appear in the auxiliary Simple Physics Simulator (SPS). The causal graph is a representational tool, not a new physical entity.

free parameters (2)
  • collision detection threshold = 23.0 units
    Used in SPS to detect collisions by distance; learned from the CLEVRER validation set (Appendix D.2). It affects counterfactual simulation results in the SOTA configuration.
  • temporal resolution t = 5 for NS-DR, 1 for IOD/VRDP
    Number of frames between positional estimates in SPS; hand-chosen based on perception availability, affects simulated trajectories.
assumptions (4)
  • domain assumption Perception is more accurate than simulation for in-video states.
    Central to the benefit of CRCG; the paper states 'as long as perception is more accurate than simulation' (Section 1) and provides evidence via Table 2.
  • domain assumption The causal graph over collision times captures all and only the causal influences relevant to the counterfactual question.
    Section 3.1; the paper notes this is an over-approximation that can fail (Remark after Definition 4).
  • domain assumption Frame-by-frame simulation model is available and reliable starting from sim nodes.
    Algorithm 1 assumes Ms can produce states at each frame; if the simulator is unreliable from the sim node, the enhanced simulation may still err.
  • standard math Answer Set Programming semantics (stable model semantics) are used correctly.
    The ASP program is provided and expected to be correct; standard logic programming background.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Think before You Simulate: Symbolic Reasoning to Orchestrate Neural Computation for Counterfactual Question Answering." pith.science (2026). https://pith.science/paper/OW5MCLMX

@misc{pith2026250610753,
  author       = {Pith},
  title        = {Pith review of: Think before You Simulate: Symbolic Reasoning to Orchestrate Neural Computation for Counterfactual Question Answering},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/OW5MCLMX}},
  note         = {Machine review of arXiv:2506.10753}
}
read the original abstract

Causal and temporal reasoning about video dynamics is a challenging problem. While neuro-symbolic models that combine symbolic reasoning with neural-based perception and prediction have shown promise, they exhibit limitations, especially in answering counterfactual questions. This paper introduces a method to enhance a neuro-symbolic model for counterfactual reasoning, leveraging symbolic reasoning about causal relations among events. We define the notion of a causal graph to represent such relations and use Answer Set Programming (ASP), a declarative logic programming method, to find how to coordinate perception and simulation modules. We validate the effectiveness of our approach on two benchmarks, CLEVRER and CRAFT. Our enhancement achieves state-of-the-art performance on the CLEVRER challenge, significantly outperforming existing models. In the case of the CRAFT benchmark, we leverage a large pre-trained language model, such as GPT-3.5 and GPT-4, as a proxy for a dynamics simulator. Our findings show that this method can further improve its performance on counterfactual questions by providing alternative prompts instructed by symbolic causal reasoning.

Figures

Figures reproduced from arXiv: 2506.10753 by the authors.

Figure 1
Figure 1. Overview of CRCG — the enhanced counterfactual reasoning using a causal graph. The Causal Graph Construction and the [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗
Figure 2
Figure 2. The causal graph for Example 1. The removed object blue is colored grey. Affected nodes are circled in red. The sim node is denoted by “s”. Definition 2 (Affected) We say a node s t o ∈ V is affected by the intervened objects Oc if (i) o ∈ Oc or (ii) there exists a node s t ′ o ′ ∈ V that is an ancestor of s t o and o ′ ∈ Oc. For any object o ∈ Ou and any frame t, if s t o is affected by Oc, the simulation for o mus… view at source ↗
Figure 3
Figure 3. Object trajectories from the location information. The [PITH_FULL_IMAGE:figures/full_fig_p005_3.png] view at source ↗
Figures from the paper (5 more)
Figure 4
Figure 4. Figure 4: Overview of the approximation of CRCG when no access to a frame-by-frame simulator is available. [PITH_FULL_IMAGE:figures/full_fig_p006_4.png]
Figure 6
Figure 6. Figure 6: Screenshots from sample CRAFT video #1,179. [PITH_FULL_IMAGE:figures/full_fig_p007_6.png]
Figure 7
Figure 7. Figure 7: Simple Physics Simulator (SPS) output for the ques [PITH_FULL_IMAGE:figures/full_fig_p014_7.png]
Figure 8
Figure 8. Figure 8: Conversion of a question and choices into ASP facts. [PITH_FULL_IMAGE:figures/full_fig_p015_8.png]
Figure 9
Figure 9. Figure 9: Conversion of improved object detection (video frame [PITH_FULL_IMAGE:figures/full_fig_p016_9.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

32 extracted references · 28 canonical work pages

  1. [1]

    Visual commonsense for scene understanding using perception, semantic parsing and reasoning

    Somak Aditya, Yezhou Yang, Chitta Baral, Cornelia Fer- muller, and Yiannis Aloimonos. Visual commonsense for scene understanding using perception, semantic parsing and reasoning. In 2015 AAAI Spring Symposium Series, 2015. 2

    work page 2015

  2. [2]

    CRAFT: A benchmark for causal reasoning about forces and interactions

    Tayfun Ates, Muhammed Samil Atesoglu, Cagatay Yigit, Ilker Kesen, Mert Kobas, Erkut Erdem, Aykut Erdem, Tilbe Goksun, and Deniz Yuret. CRAFT: A benchmark for causal reasoning about forces and interactions. arXiv preprint arXiv:2012.04293, 2020. 2, 7, 8, 11

    arXiv 2012

  3. [3]

    Cophy: Counterfactual learning of phys- ical dynamics

    Fabien Baradel, Natalia Neverova, Julien Mille, Greg Mori, and Christian Wolf. Cophy: Counterfactual learning of phys- ical dynamics. In International Conference on Learning Representations, 2020. 8

    work page 2020

  4. [4]

    Answer set programming at a glance

    Gerhard Brewka, Ilkka Niemel ¨a, and Miroslaw Truszczyn- ski. Answer set programming at a glance. Communications of the ACM, 54(12):92–103, 2011. 2

    work page 2011

  5. [5]

    Lan- guage models are few-shot learners

    Tom Brown, Benjamin Mann, Nick Ryder, Melanie Sub- biah, Jared D Kaplan, Prafulla Dhariwal, Arvind Neelakan- tan, Pranav Shyam, Girish Sastry, Amanda Askell, et al. Lan- guage models are few-shot learners. Advances in neural in- formation processing systems, 33:1877–1901, 2020. 2

    1901

  6. [6]

    Ground- ing physical concepts of objects and events through dynamic visual reasoning

    Zhenfang Chen, Jiayuan Mao, Jiajun Wu, Kwan-Yee Ken- neth Wong, Joshua B Tenenbaum, and Chuang Gan. Ground- ing physical concepts of objects and events through dynamic visual reasoning. In ICLR, 2021. 2, 6, 7

    work page 2021

  7. [7]

    Tenenbaum, and Chuang Gan

    Zhenfang Chen, Kexin Yi, Yunzhu Li, Mingyu Ding, Anto- nio Torralba, Joshua B. Tenenbaum, and Chuang Gan. Com- phy: Compositional physical reasoning of objects and events from videos. In International Conference on Learning Rep- resentations, 2022. 8

    work page 2022

  8. [8]

    Attention over learned object embed- dings enables complex visual reasoning

    David Ding, Felix Hill, Adam Santoro, Malcolm Reynolds, and Matt Botvinick. Attention over learned object embed- dings enables complex visual reasoning. Advances in neural information processing systems, 34:9112–9124, 2021. 1, 6, 7

    work page 2021

Show all 32 references
  1. [9]

    Dynamic visual reasoning by learning differentiable physics models from video and lan- guage

    Mingyu Ding, Zhenfang Chen, Tao Du, Ping Luo, Josh Tenenbaum, and Chuang Gan. Dynamic visual reasoning by learning differentiable physics models from video and lan- guage. Advances in Neural Information Processing Systems, 34, 2021. 1, 2, 6, 7

    2021

  2. [10]

    The stable model semantics for logic programming

    Michael Gelfond and Vladimir Lifschitz. The stable model semantics for logic programming. In Robert Kowalski and Kenneth Bowen, editors, Proceedings of International Logic Programming Conference and Symposium , pages 1070–

  3. [11]

    Seeing out of the box: End-to- end pre-training for vision-language representation learning

    Zhicheng Huang, Zhaoyang Zeng, Yupan Huang, Bei Liu, Dongmei Fu, and Jianlong Fu. Seeing out of the box: End-to- end pre-training for vision-language representation learning. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 12976–1298...

    2021

  4. [12]

    Compositional attention networks for machine reasoning

    Drew A Hudson and Christopher D Manning. Compositional attention networks for machine reasoning. arXiv preprint arXiv:1803.03067, 2018. 1

    2018 arXiv

  5. [13]

    Inferring and executing programs for visual reasoning

    Justin Johnson, Bharath Hariharan, Laurens Van Der Maaten, Judy Hoffman, Li Fei-Fei, C Lawrence Zitnick, and Ross Girshick. Inferring and executing programs for visual reasoning. In Proceedings of the IEEE International Conference on Computer Vision, pages 2989–2998, 2017. 1

    2017

  6. [14]

    Visual reasoning on complex events in soccer videos using answer set programming

    Abdullah Khan, Loris Bozzato, Luciano Serafini, and Beat- rice Lazzerini. Visual reasoning on complex events in soccer videos using answer set programming. In GCAI, pages 42– 53, 2019. 2

    2019

  7. [15]

    What is answer set programming? In Proceedings of the AAAI Conference on Artificial Intelli- gence, pages 1594–1597

    Vladimir Lifschitz. What is answer set programming? In Proceedings of the AAAI Conference on Artificial Intelli- gence, pages 1594–1597. MIT Press, 2008. 2

    2008

  8. [16]

    Answer set programming

    Vladimir Lifschitz. Answer set programming . Springer,

  9. [17]

    Deepproblog: Neu- ral probabilistic logic programming

    Robin Manhaeve, Sebastijan Dumancic, Angelika Kimmig, Thomas Demeester, and Luc De Raedt. Deepproblog: Neu- ral probabilistic logic programming. In Proceedings of Ad- vances in Neural Information Processing Systems , pages 3749–3759, 2018

    2018

  10. [18]

    Tenenbaum, and Jiajun Wu

    Jiayuan Mao, Chuang Gan, Pushmeet Kohli, Joshua B. Tenenbaum, and Jiajun Wu. The neuro-symbolic concept learner: interpreting scenes, words, and sentences from nat- ural supervision. In ICLR, 2019. 7

    2019

  11. [19]

    Transparency by design: Closing the gap be- tween performance and interpretability in visual reasoning

    David Mascharka, Philip Tran, Ryan Soklaski, and Arjun Majumdar. Transparency by design: Closing the gap be- tween performance and interpretability in visual reasoning. In Proceedings of the IEEE conference on computer vision and pattern recognition, pages 4942–4950, 2018. 1

    2018

  12. [20]

    Adavit: Adaptive vision transformers for efficient image recognition

    Lingchen Meng, Hengduo Li, Bor-Chun Chen, Shiyi Lan, Zuxuan Wu, Yu-Gang Jiang, and Ser-Nam Lim. Adavit: Adaptive vision transformers for efficient image recognition. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 12309–12318, 2022. 1

    2022

  13. [21]

    Gpt-4 technical report, 2023

    OpenAI. Gpt-4 technical report, 2023. 2

    2023

  14. [22]

    Clevr hyp: A challenge dataset and baselines for visual question answering with hypothetical actions over images

    Shailaja Keyur Sampat, Akshay Kumar, Yezhou Yang, and Chitta Baral. Clevr hyp: A challenge dataset and baselines for visual question answering with hypothetical actions over images. In Proceedings of the 2021 Conference of the North American Chapter of the Association for Comp...

    2021

  15. [23]

    Hyster: A hybrid spatio-temporal event reasoner

    Theophile Sautory, Nuri Cingillioglu, and Alessandra Russo. Hyster: A hybrid spatio-temporal event reasoner. arXiv preprint arXiv:2101.06644, 2021. 2

    2021 arXiv

  16. [24]

    Learning from simulated and unsupervised images through adversarial training

    Ashish Shrivastava, Tomas Pfister, Oncel Tuzel, Joshua Susskind, Wenda Wang, and Russell Webb. Learning from simulated and unsupervised images through adversarial training. In Proceedings of the IEEE conference on computer vision and pattern recognition, pages 2107–2116, 2017. 1

    2017

  17. [25]

    Learning from noisy labels with deep neural networks: A survey

    Hwanjun Song, Minseok Kim, Dongmin Park, Yooju Shin, and Jae-Gil Lee. Learning from noisy labels with deep neural networks: A survey. IEEE Transactions on Neural Networks and Learning Systems, 2022. 1

    2022

  18. [26]

    Out of sight but not out of mind: An answer set programming based online abduction framework for visual sensemaking in autonomous driving

    Jakob Suchan, Mehul Bhatt, and Srikrishna Varadarajan. Out of sight but not out of mind: An answer set programming based online abduction framework for visual sensemaking in autonomous driving. In 28th International Joint Conference on Artificial Intelligence (IJCAI 2019), Mac...

    2019

  19. [27]

    Visual explanation by high-level abduction: On answer-set programming driven reasoning about moving ob- jects

    Jakob Suchan, Mehul Bhatt, Przemysław Wałega, and Carl Schultz. Visual explanation by high-level abduction: On answer-set programming driven reasoning about moving ob- jects. In Proceedings of the AAAI Conference on Artificial Intelligence, volume 32, 2018. 2

    2018

  20. [28]

    Object dynamics distillation for scene decomposition and representation

    Qu Tang, Xiangyu Zhu, Zhen Lei, and Zhaoxiang Zhang. Object dynamics distillation for scene decomposition and representation. In International Conference on Learning Representations, 2022. 1, 6, 7

    2022

  21. [29]

    Mind games: Game engines as an ar- chitecture for intuitive physics

    Tomer D Ullman, Elizabeth Spelke, Peter Battaglia, and Joshua B Tenenbaum. Mind games: Game engines as an ar- chitecture for intuitive physics. Trends in cognitive sciences, 21(9):649–665, 2017. 1

    2017

  22. [30]

    An- swering visual what-if questions: From actions to predicted scene descriptions

    Misha Wagner, Hector Basevi, Rakshith Shetty, Wenbin Li, Mateusz Malinowski, Mario Fritz, and Ales Leonardis. An- swering visual what-if questions: From actions to predicted scene descriptions. In Proceedings of the European Con- ference on Computer Vision (ECCV) Workshops, pages 0–0,

  23. [31]

    NeurASP: Em- bracing neural networks into answer set programming

    Zhun Yang, Adam Ishay, and Joohyung Lee. NeurASP: Em- bracing neural networks into answer set programming. In Proceedings of International Joint Conference on Artificial Intelligence (IJCAI), pages 1755–1762, 2020

    2020

  24. [32]

    Think before You Simu- late: Symbolic Reasoning to Orchestrate Neural Computation for Counterfactual Question Answering

    Kexin Yi, Chuang Gan, Yunzhu Li, Pushmeet Kohli, Ji- ajun Wu, Antonio Torralba, and Joshua B Tenenbaum. CLEVRER: Collision events for video representation and reasoning. In ICLR, 2019. 1, 6, 11 Appendix to “Think before You Simu- late: Symbolic Reasoning to Orchestrate Neural ...

    2019

Pith tools

Reviewed August 7, 2026 · model on record in the stance chip above.