# Automatically Composing Representation Transformations as a Means for Generalization

arXiv: https://arxiv.org/abs/1807.04640 · HTML: https://mlsanity.com/paper/1807.04640/

## Key facts

- Cited by 35 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 12 Jul 2018.

## Built on by (most cited first, 30 of 35)

- [Multi-Task Reinforcement Learning with Context-based Representations](https://mlsanity.com/paper/2102.06177/) (2021)
- [Modular Deep Learning](https://mlsanity.com/paper/2302.11529/) (2023)
- [Can Subnetwork Structure be the Key to Out-of-Distribution Generalization?](https://mlsanity.com/paper/2106.02890/) (2021)
- [Wide Neural Networks Forget Less Catastrophically](https://mlsanity.com/paper/2110.11526/) (2021)
- [Continual Learning via Local Module Composition](https://mlsanity.com/paper/2111.07736/) (2021)
- [Modular Lifelong Reinforcement Learning via Neural Composition](https://mlsanity.com/paper/2207.00429/) (2022)
- [Inference-Time Policy Adapters (IPA): Tailoring Extreme-Scale LMs without Fine-tuning](https://mlsanity.com/paper/2305.15065/) (2023)
- [Continual Learning with Recursive Gradient Optimization](https://mlsanity.com/paper/2201.12522/) (2022)
- [Position: Theory of Mind Benchmarks are Broken for Large Language Models](https://mlsanity.com/paper/2412.19726/) (2024)
- [How to Reuse and Compose Knowledge for a Lifetime of Tasks: A Survey on Continual Learning and Functional Composition](https://mlsanity.com/paper/2207.07730/) (2022)
- [Dynamic Inference with Neural Interpreters](https://mlsanity.com/paper/2110.06399/) (2021)
- [CompoSuite: A Compositional Reinforcement Learning Benchmark](https://mlsanity.com/paper/2207.04136/) (2022)
- [Solving Compositional Reinforcement Learning Problems via Task Reduction](https://mlsanity.com/paper/2103.07607/) (2021)
- [A Survey of Continual Reinforcement Learning](https://mlsanity.com/paper/2506.21872/) (2025)
- [Decrypting Cryptic Crosswords: Semantically Complex Wordplay Puzzles as a Target for NLP](https://mlsanity.com/paper/2104.08620/) (2021)
- [Neural Relational Inference with Fast Modular Meta-learning](https://mlsanity.com/paper/2310.07015/) (2023)
- [On The Specialization of Neural Modules](https://mlsanity.com/paper/2409.14981/) (2024)
- [Revisiting Replay and Gradient Alignment for Continual Pre-Training of Large Language Models](https://mlsanity.com/paper/2508.01908/) (2025)
- [REFACTOR: Learning to Extract Theorems from Proofs](https://mlsanity.com/paper/2402.17032/) (2024)
- [Enabling Realtime Reinforcement Learning at Scale with Staggered Asynchronous Inference](https://mlsanity.com/paper/2412.14355/) (2024)
- [Modularity in Reinforcement Learning via Algorithmic Independence in Credit Assignment](https://mlsanity.com/paper/2106.14993/) (2021)
- [A Theory of Initialisation's Impact on Specialisation](https://mlsanity.com/paper/2503.02526/) (2025)
- [Breaking Neural Network Scaling Laws with Modularity](https://mlsanity.com/paper/2409.05780/) (2024)
- [Gradient-Based Program Synthesis with Neurally Interpreted Languages](https://mlsanity.com/paper/2604.18907/) (2026)
- [DAIR: Disentangled Attention Intrinsic Regularization for Safe and Efficient Bimanual Manipulation](https://mlsanity.com/paper/2106.05907/) (2021)
- [Modularity in Deep Learning: A Survey](https://mlsanity.com/paper/2310.01154/) (2023)
- [Beyond Sliding Windows: Learning to Manage Memory in Non-Markovian Environments](https://mlsanity.com/paper/2512.19154/) (2025)
- [Compositional meta-learning through probabilistic task inference](https://mlsanity.com/paper/2510.01858/) (2025)
- [The Effectiveness of Approximate Regularized Replay for Efficient Supervised Fine-Tuning of Large Language Models](https://mlsanity.com/paper/2512.22337/) (2025)
- [Modularity is the Bedrock of Natural and Artificial Intelligence](https://mlsanity.com/paper/2602.18960/) (2026)

## Abstract (arXiv)

A generally intelligent learner should generalize to more complex tasks than it has previously encountered, but the two common paradigms in machine learning -- either training a separate learner per task or training a single learner for all tasks -- both have difficulty with such generalization because they do not leverage the compositional structure of the task distribution. This paper introduces the compositional problem graph as a broadly applicable formalism to relate tasks of different complexity in terms of problems with shared subproblems. We propose the compositional generalization problem for measuring how readily old knowledge can be reused and hence built upon. As a first step for tackling compositional generalization, we introduce the compositional recursive learner, a domain-general framework for learning algorithmic procedures for composing representation transformations, producing a learner that reasons about what computation to execute by making analogies to previously seen problems. We show on a symbolic and a high-dimensional domain that our compositional approach can generalize to more complex problems than the learner has previously encountered, whereas baselines that are not explicitly compositional do not.