mlsanityData to 7 Oct 2026

Compositional Attention Networks for Machine Reasoning

arXiv 1803.03067 ↗First posted 8 Mar 2018cs.AIcited by 183

Not on the map. Why?

Cited by 32 in LLM reasoning, but placed by its own abstract, which didn’t fit.

Abstract (arXiv)

We present the MAC network, a novel fully differentiable neural network architecture, designed to facilitate explicit and expressive reasoning. MAC moves away from monolithic black-box neural architectures towards a design that encourages both transparency and versatility. The model approaches problems by decomposing them into a series of attention-based reasoning steps, each performed by a novel recurrent Memory, Attention, and Composition (MAC) cell that maintains a separation between control and memory. By stringing the cells together and imposing structural constraints that regulate their interaction, MAC effectively learns to perform iterative reasoning processes that are directly inferred from the data in an end-to-end approach. We demonstrate the model's strength, robustness and interpretability on the challenging CLEVR dataset for visual reasoning, achieving a new state-of-the-art 98.9% accuracy, halving the error rate of the previous best model. More importantly, we show that the model is computationally-efficient and data-efficient, in particular requiring 5x less data than existing models to achieve strong results.

Key facts. Cited by 183 papers in the map (data through 7 Oct 2026). First posted on arXiv 8 Mar 2018.

Builds on

No reference list yet.

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2021: 54; 2022: 53; 2023: 33; 2024: 16; 2025: 19; 2026: 8.

183 papers.

Sort:

By area:

  1. 2021ViLT: Vision-and-Language Transformer Without Convolution or Region Supervisioncited by 1,224
  2. 2021MDETR -- Modulated Detection for End-to-End Multi-Modal Understandingcited by 811
  3. 2023ViperGPT: Visual Inference via Python Execution for Reasoningcited by 591
  4. 2022CrossPoint: Self-Supervised Cross-Modal Contrastive Learning for 3D Point Cloud Understandingcited by 198
  5. 2022CREPE: Can Vision-Language Foundation Models Reason Compositionally?cited by 177
  6. 2021AGQA: A Benchmark for Compositional Spatio-Temporal Reasoningcited by 120
  7. 2022Foundations and Trends in Multimodal Machine Learning: Principles, Challenges, and Open Questionscited by 116
  8. 2021Inducing Causal Structure for Interpretable Neural Networkscited by 107
  9. 2021Improving Multi-hop Knowledge Base Question Answering by Learning Intermediate Supervision Signalscited by 100
  10. 2021Open World Compositional Zero-Shot Learningcited by 100
  11. 2022Benchmarking Spatial Relationships in Text-to-Image Generationcited by 96
  12. 2022Super-CLEVR: A Virtual Benchmark to Diagnose Domain Robustness in Visual Reasoningcited by 90
38 more
  1. 2025Perception, Reason, Think, and Plan: A Survey on Large Multimodal Reasoning Modelscited by 89
  2. 2021SUTD-TrafficQA: A Question Answering Benchmark and an Efficient Network for Video Reasoning over Traffic Eventscited by 84
  3. 2024Understanding the Limits of Vision Language Models Through the Lens of the Binding Problemcited by 71
  4. 2021Grounding Physical Concepts of Objects and Events Through Dynamic Visual Reasoningcited by 68
  5. 2021Probing Inter-modality: Visual Parsing with Self-Attention for Vision-Language Pre-trainingcited by 67
  6. 2022ComPhy: Compositional Physical Reasoning of Objects and Events from Videoscited by 61
  7. 20233D Concept Learning and Reasoning from Multi-View Imagescited by 60
  8. 2023Coupling Large Language Models with Logic Programming for Robust and General Reasoning from Textcited by 59
  9. 2021Dynamic Visual Reasoning by Learning Differentiable Physics Models from Video and Languagecited by 55
  10. 2023EarthVQA: Towards Queryable Earth via Relational Reasoning-Based Remote Sensing Visual Question Answeringcited by 51
  11. 2021Attention, please! A survey of Neural Attention Models in Deep Learningcited by 50
  12. 2021A Survey on Neural-symbolic Learning Systemscited by 39
  13. 2022Modular Lifelong Reinforcement Learning via Neural Compositioncited by 39
  14. 2021PTR: A Benchmark for Part-based Conceptual, Relational, and Physical Reasoningcited by 34
  15. 2022Towards Data-and Knowledge-Driven Artificial Intelligence: A Survey on Neuro-Symbolic Computingcited by 34
  16. 2023What's Left? Concept Grounding with Logic-Enhanced Foundation Modelscited by 33
  17. 2023GPT-4 Enhanced Multimodal Grounding for Autonomous Driving: Leveraging Cross-Modal Attention with Large Language Modelscited by 31
  18. 2021Automatic Generation of Contrast Sets from Scene Graphs: Probing the Compositional Consistency of GQAcited by 30
  19. 2023LLark: A Multimodal Instruction-Following Language Model for Musiccited by 29
  20. 2022Vision-Language Intelligence: Tasks, Representation Learning, and Large Modelscited by 28
  21. 2022On Neural Architecture Inductive Biases for Relational Taskscited by 26
  22. 2021COVR: A test-bed for Visually Grounded Compositional Generalization with real imagescited by 25
  23. 2021Separating Skills and Concepts for Novel Visual Question Answeringcited by 24
  24. 2021Dynamic Inference with Neural Interpreterscited by 24
  25. 2021Found a Reason for me? Weakly-supervised Grounded Visual Question Answering using Capsulescited by 22
  26. 2021Neuro-Symbolic AI: An Emerging Class of AI Workloads and their Characterizationcited by 22
  27. 2024Remote Sensing SpatioTemporal Vision-Language Models: A Comprehensive Surveycited by 22
  28. 2021Iterated learning for emergent systematicity in VQAcited by 21
  29. 2022Towards Lightweight Transformer via Group-wise Transformation for Vision-and-Language Taskscited by 21
  30. 2022CompoSuite: A Compositional Reinforcement Learning Benchmarkcited by 21
  31. 2021Understanding the Role of Scene Graphs in Visual Question Answeringcited by 20
  32. 2022FALCON: Fast Visual Concept Learning by Integrating Images, Linguistic descriptions, and Conceptual Relationscited by 20
  33. 2021ReaSCAN: Compositional Reasoning in Language Groundingcited by 19
  34. 2021ProTo: Program-Guided Transformer for Program-Guided Taskscited by 19
  35. 2021COMPOSER: Compositional Reasoning of Group Activity in Videos with Keypoint-Only Modalitycited by 19
  36. 2022V-Doc : Visual questions answers with Documentscited by 19
  37. 2023Variational Cross-Graph Reasoning and Adaptive Structured Semantics Learning for Compositional Temporal Groundingcited by 19
  38. 2023Improved Logical Reasoning of Language Models via Differentiable Symbolic Programmingcited by 18

50 most cited of 183 All 183 (JSON)