mlsanityData to 7 Oct 2026

OC-NMN: Object-centric Compositional Neural Module Network for Generative Visual Analogical Reasoning

arXiv 2310.18807 ↗First posted 28 Oct 2023cs.AI cs.CVcited by 0

On the map: LLM reasoning → Multimodal reasoning

Abstract (arXiv)

A key aspect of human intelligence is the ability to imagine -- composing learned concepts in novel ways -- to make sense of new scenarios. Such capacity is not yet attained for machine learning systems. In this work, in the context of visual reasoning, we show how modularity can be leveraged to derive a compositional data augmentation framework inspired by imagination. Our method, denoted Object-centric Compositional Neural Module Network (OC-NMN), decomposes visual generative reasoning tasks into a series of primitives applied to objects without using a domain-specific language. We show that our modular architectural choices can be used to generate new training tasks that lead to better out-of-distribution generalization. We compare our model to existing and new baselines in proposed visual reasoning benchmark that consists of applying arithmetic operations to MNIST digits.

Key facts. No citing papers in the map yet (data through 7 Oct 2026). Placed in Multimodal reasoning, one of 17 directions in LLM reasoning. Builds on 37 of its 54 references that are in the map. First posted on arXiv 28 Oct 2023.

Builds on

37 of 54 references.

References · 37

  1. 2022Scaling Instruction-Finetuned Language Modelscited by 2,997
  2. 2019Invariant Risk Minimizationcited by 1,832
  3. 2019Dream to Control: Learning Behaviors by Latent Imaginationcited by 1,598
  4. 2018ImageNet-trained CNNs are biased towards texture; increasing shape bias improves accuracy and robustnesscited by 1,430
  5. 2020Mastering Atari with Discrete World Modelscited by 990
  6. 2020Object-Centric Learning with Slot Attentioncited by 865
  7. 2017One pixel attack for fooling deep neural networkscited by 579
  8. 2019MONet: Unsupervised Scene Decomposition and Representationcited by 388
  9. 2015Neural Module Networkscited by 375
  10. 2019Multi-Object Representation Learning with Iterative Variational Inferencecited by 359
  11. 2020Inductive Biases for Deep Learning of Higher-Level Cognitioncited by 273
  12. 2022Emergent Analogical Reasoning in Large Language Modelscited by 239
25 more
  1. 2019RAVEN: A Dataset for Relational and Analogical Visual rEasoNingcited by 225
  2. 2019Recurrent Independent Mechanismscited by 219
  3. 2016Attend, Infer, Repeat: Fast Scene Understanding with Generative Modelscited by 209
  4. 2018Measuring abstract reasoning in neural networkscited by 165
  5. 2019A Meta-Transfer Objective for Learning to Disentangle Causal Mechanismscited by 158
  6. 2018Sequential Attend, Infer, Repeat: Generative Modelling of Moving Objectscited by 128
  7. 2017Neural Expectation Maximizationcited by 123
  8. 2017Learning Independent Causal Mechanismscited by 104
  9. 2021Abstraction and Analogy-Making in Artificial Intelligencecited by 103
  10. 2018Relational Neural Expectation Maximization: Unsupervised Discovery of Objects and their Interactionscited by 100
  11. 2023The ConceptARC Benchmark: Evaluating Understanding and Generalization in the ARC Domaincited by 97
  12. 2015Neural Programmer-Interpreterscited by 96
  13. 2021Neural Production Systems: Learning Rule-Governed Visual Dynamicscited by 75
  14. 2020Improving Generative Imagination in Object-Centric World Modelscited by 70
  15. 2019CLOSURE: Assessing Systematic Generalization of CLEVR Modelscited by 63
  16. 2021Continual Learning via Local Module Compositioncited by 61
  17. 2020Closed Loop Neural-Symbolic Learning via Integrating Neural Perception, Grammar Parsing, and Symbolic Reasoningcited by 58
  18. 2020Object Files and Schemata: Factorizing Declarative and Procedural Knowledge in Dynamical Systemscited by 48
  19. 2020Emergent Symbols through Binding in External Memorycited by 47
  20. 2017Making Neural Programming Architectures Generalize via Recursioncited by 35
  21. 2017IQ of Neural Networkscited by 16
  22. 2020The Overfitted Brain: Dreams evolved to assist generalizationcited by 16
  23. 2019A Perspective on Objects and Systematic Generalization in Model-Based RLcited by 14
  24. 2022Improving Systematic Generalization Through Modularity and Augmentationcited by 13
  25. 2020Neural Abstract Reasonercited by 7

Built on by

Not cited yet.

Similar, not cited

Close in wording, no citation either way

  1. 2025A Neurosymbolic Agent System for Compositional Visual Reasoning
  2. 2022A Benchmark for Compositional Visual Reasoning
  3. 2025COGITAO: A Visual Reasoning Framework To Study Compositionality & Generalization
  4. 2024Learning to Compose: Improving Object Centric Learning by Injecting Compositionality
  5. 2022Visual Programming: Compositional visual reasoning without training
  6. 2023Im-Promptu: In-Context Composition from Image Prompts
  7. 2025NePTune: A Neuro-Pythonic Framework for Tunable Compositional Reasoning on Vision-Language
  8. 2026Compositional Neuro-Symbolic Reasoning