Compositional Generalization

Follow this topic Ask about this topic

Compositional generalization studies whether models can reuse learned concepts, skills, and rules in combinations they did not encounter during training. Approaches include modular and object-centric architectures, factorized or symbolic representations, and training methods such as curricula, structured reasoning traces, and reinforcement learning. Researchers test these abilities on controlled language, vision, arithmetic, planning, and robotics tasks, often holding out combinations, deeper structures, or longer sequences. Results remain uneven: models often recombine familiar parts more reliably than they discover new ones, and strong test accuracy can reflect shortcuts rather than reusable structure.

Latest papers

113 papers12 papers in the last four weeks, up 9% on the four weeks before. 0.1% of all new papers.

Quantum Models with Multi-Stage Training for Compositional Concept Generalization
Aug 16, 2026Mina Abbaszadeh, Matilda Karabina Moore, Mehrnoosh Sadrzadeh +1Compositional Generalization Compositional Representation Learning
Department of Computer Science University College London London, United Kingdom · Institute for Logic, Language and Computation University of Amsterdam Amsterdam, Netherlands
Compositional Benchmark Synthesis for Hierarchical Human Action Recognition
Aug 11, 2026Farnaz Soleimani, Abdelghani Chibani, Yacine Amirat +1Human Activity Recognition Synthetic Benchmark Generation
LISSI
Support Operation Factorization: Compositional Readout of Frozen Vision Encoders under Controlled Interventions
Aug 6, 2026Zhongyao Wang, Wanli Ouyang, Taoyong Cui +1Visual Representation Learning Representation Disentanglement
Fudan University · MMLab, The Chinese University of Hong Kong · Department of Computer Science and Engineering, The Chinese University of Hong Kong
Evolving language compositionality in a frequency-structured meaning space
Jul 31, 2026Fabio De Ponte, Eloise Gaines-White, Conor Houghton +1Compositional Generalization
Université de Namur and Vrije Universiteit Brussel, Belgium · University of Bristol, UK
Scale Up Strategically: Learning Compositional Generalization via Bias-Aware Evaluation and Data Collection for Robotic Manipulation
Jul 23, 2026Yu Qi, Zhang Ye, Xinyi Xu +6Robot Skill Learning Language-Conditioned Robot Manipulation
Northeastern University · NVIDIA
Can Generative Recommendation Reach Cold Items? A Temporal Perspective on Semantic-ID Generation
Jul 23, 2026Jie Peng, Yanping Zheng, Zhewei Zhe +3Semantic ID Recommendation Sequential Recommendation
Renmin University of China · Beijing, China · Alibaba
On the Computational Complexity of Structural Generalization
Jul 21, 2026Zichao WeiTransformer Expressivity Compositional Generalization
Department of Language Science and Technology Saarland University Saarbrücken, Germany
EditCLEVR: A Paired-Scene Intervention Benchmark for Compositional Faithfulness of Object-Centric Representations
Jul 19, 2026Anuraag Gadehothur Karnam, Tarunesh SathishCounterfactual Evaluation Object-Centric Representation Learning
Lebanon Trail High School · Walnut Grove High School
On Locality and Length Generalization in Visual Reasoning
Jul 10, 2026Pulkit Madan, Sanjay Haresh, Reza Ebrahimi +3Visual Reasoning Length Generalization
Qualcomm AI Research*
Progressive Reasoning with Primitive Correction for Compositional Zero-Shot Learning
Jul 7, 2026Ziyi Chen, Haoyan Shi, Sunhan Xu +1Zero-Shot Learning VLM Reasoning
School of Computer Science and Technology, Beijing Jiaotong University · Key Laboratory of Big Data & Artificial Intelligence in Transportation (Ministry of Education)
ClassicLogic: A Knowledge-Driven Benchmark of Classic Puzzle Games for Evaluating Compositional Generalization
Jul 6, 2026Mahnoor Shahid, Hannes RotheLogical Reasoning Benchmark Design
Universität Duisburg-Essen Germany
InternVLA-A1.5: Unifying Understanding, Latent Foresight, and Action for Compositional Generalization
Jul 6, 2026Haoxiang Ma, Junhao Cai, Xiaoxu Xu +26Generalization in Robotic Manipulation Visuomotor Policy Learning
Physical Intelligence Team, Shanghai AI Laboratory
Input Pathways Shape Few-Shot, Not Zero-Shot, Binding in Tiny Transformers: A Fully-Enumerable Study
Jul 6, 2026Yoshiyuki OotaniRepresentation Learning Transformer
Independent Researcher
Open-Set Source Tracing as Compositional Factors via Structured Prototypes
Jul 3, 2026Santiago Rubio, Antonio Almudévar, Antonio Miguel +2Audio Deepfake Detection Prototype Learning
ViVoLab, I3A University of Zaragoza Zaragoza, Spain
On the Role of Directionality in Structural Generalization
Jul 2, 2026Zichao WeiSemantic Parsing Compositional Generalization
Saarland University
Learning to Reason with Curriculum II: Compositional Generalization
Jun 26, 2026Nived Rajaraman, Audrey Huang, Miroslav Dudik +3Curriculum Learning Automata Learning
Microsoft Research. · University of Illinois Urbana-Champaign. · Microsoft Research, New York, NY.
Empowering GUI Agents via Autonomous Experience Exploration and Hindsight Experience Utilization for Task Planning
Jun 25, 2026Tianyi Men, Zhuoran Jin, Pengfei Cao +3GUI Agents LLM Agent Training
The Key Laboratory of Cognition and Decision Intelligence for Complex Systems, Institute of Automation, Chinese Academy of Sciences, Beijing, China · School of Artificial Intelligence, University of Chinese Academy of Sciences, Beijing, China
Project Auto-World: Towards Automated Benchmarking of Neural Relational Reasoners
Jun 23, 2026Anirban Das, Joanne Boisson, Irtaza Khalid +2Language Model-Guided Search Heuristics Relational Reasoning
Cardiff University · University of Pennsylvania
Catastrophic Compositional Generation: Why Vanilla Diffusion Models Fail to Extrapolate
Jun 22, 2026Duncan Soiffer, Chandler Squires, Yuan Guan +2OOD Generalization Conditional Diffusion Models
Machine Learning Department, Carnegie Mellon University · Valence Labs · Department of Computer Science, University of Manchester
Compositionality Emerges in a Narrow Depth-Connectivity Regime: Architecture Constraints and Solution Manifolds
Jun 18, 2026Dat H. Do, Rushi Shah, Duc V. Le +1Sparse Neural Networks Compositional Representation Learning
National University of Singapore1 · University of Twente2
From Reasoning Traces to Reusable Modules: Understanding Compositional Generalization in Language Model Reasoning
Jun 16, 2026Lingjing Kong, Xin Liu, Guangyi Chen +9LLM Fine-Tuning RL for Language Model Reasoning
Carnegie Mellon University · Mohamed bin Zayed University of Artificial Intelligence · Institute of Foundation Models +1
A Compositional Framework for Open-ended Intelligence
Jun 13, 2026Ida Momennejad, Roberta RaileanuOpen-Ended Generation Compositional Generalization
Microsoft Research NYC, New York, USA · Google DeepMind, London, UK.
Fodor and Pylyshyn's Systematicity Challenge Still Stands
Jun 12, 2026Michael Goodale, Salvador MascarenhasComputational Cognitive Modeling Compositional Generalization
Institut Jean Nicod, Département d’études cognitives · ENS, EHESS, CNRS, PSL University.
Adversarial Concept Search: Predicting Compositional Errors From Feature Geometry
Jun 11, 2026Jennifer Meng Lu, Ruochen Zhang, Isabelle Lee +3LLM Interpretability Representation Geometry in Language Models
Brown University · University of Southern California · Harvard University +1
Compositional Generative Modeling from Decentralized Data
Jun 8, 2026Mashrur M. Morshed, Vishnu Naresh BoddetiFlow Matching Decentralized Learning
Department of Computer Science and Engineering, Michigan State University, East Lansing, MI, USA.
Disentanglement-Based Equivariant Learning for Compositional VQA
Jun 1, 2026Zhou Du, Zhaoquan Yuan, Xiao Wu +1Visual Question Answering Disentangled Representation Learning
School of Computing and Artificial Intelligence, Southwest Jiaotong University, Chengdu 611730, China, and also with Engineering Research Center of Sustainable Urban Intelligence Transportation, Ministry of Education, China · State Key Laboratory of Multimodal Artificial Intelligence Systems (MAIS), Institute of Automation, Chinese Academy of Sciences, Beijing 100190, China, and with the School of Artificial Intelligence, University of Chinese Academy of Sciences, Beijing 101408, China, and also with Peng Cheng Laboratory, ShenZhen 518055, China
How can embedding models bind concepts?
May 29, 2026Arnas Uselis, Darina Koishigarina, Seong Joon OhMultimodal Embedding Vision-Language Alignment
University of Tübingen, Tübingen · 2KAIST AI
Assign and Add: A Mechanistic Study of Compositional Arithmetic
May 29, 2026Brady Exoo, Alberto Bietti, John SousTransformer Interpretability Transformer
Yale University · Flatiron Institute
Building Generalization Into Behavior Generation Via Adaptive Compositions of Regularities
May 29, 2026Aravind Battaje, Malte Bernhard, Vito Mengers +1Robot Motion Generation Robot Policy Generalization
Robotics and Biology Laboratory, Technische Universität Berlin, Germany · Science of Intelligence, Research Cluster of Excellence, Berlin, Germany · Robotics Institute Germany
Unveiling the Visual Counting Bottleneck in Vision-Language Models
May 28, 2026Xingzhou Pang, Yifan Hou, Junling Wang +1Object Counting Latent Visual Reasoning
Department of Computer Science, ETH Zürich

Aug 16, 2026J/K move · Enter open · S save

Quantum Models with Multi-Stage Training for Compositional Concept Generalization

Mina Abbaszadeh, Matilda Karabina Moore, Mehrnoosh Sadrzadeh, Martha Lewis

Department of Computer Science University College London London, United Kingdom · Institute for Logic, Language and Computation University of Amsterdam Amsterdam, Netherlands

Compositional Generalization Compositional Representation Learning Quantum Machine Learning

Open paper AskPDF arXiv

Compositional Concept Generalization (CoCoGen), the ability to systematically recombine learned primitives in novel contexts, is a key challenge for multimodal learning. In this work, we provide a solution using a compositional model of meaning that separates nouns from relations and uses tensors and variational quantum circuits to train them on data. This model enables us to employ a multi stage training paradigm, one that first learns object representations from single-object image-caption pairs, then subsequently transfers these to the relational stage where object parameters are frozen and optimisation is only applied to relational components. This design explicitly enforces compositional factorisation at the circuit, ensuring that relations are learned as transformations over stable primitives. The training paradigm is tested on the CLEVR dataset developed specificially for CoCoGen. For text, we work with vector representations of nouns and higher order tensor representations of relations using a set of different ansatz. For images, we work with quantum encodings of image embeddings dervied from Open AI's Vision Language tool CLIP and contrast amplitude encoding, which preserves the original embedding geometry, with angle encoding, which introduces nonlinear feature transformations. Our results show that multi-staged training combined with structured encodings significantly improves out of distribution relational generalisation, while using orders of magnitude fewer trainable parameters than classical baselines. We find that performance gains arise from the interaction between representation and encoding, with nonlinear quantum encodings enhancing the separability of compositional structure. These findings demonstrate that structured quantum representations and staged learning provide an effective framework for compositional generalisation in multimodal quantum machine learning.