RECORD
Concepts, Properties and an Approach for Compositional Generalization
Artificial Intelligence · Submitted 8 Feb 2021
abstract compiled · 2026-09-23 · no language-model review
The authors' main claim We hope this work would be helpful to clarify fundamentals of compositional generalization and lead to advance artificial intelligence.
The compiled record
The paper's own sentences, arranged. Each label is drawn from a closed vocabulary; the number beside it is the probability the classifier gave that label. Numbers in the tables are the authors', checked by code. No sentence in this record was written by a model.
Open the compiled record
The abstract, sentence by sentence
Compositional generalization is the capacity to recognize and imagine a large amount of novel combinations from known components. It is a key in human intelligence, but current neural networks generally lack such ability. This report connects a series of our work for compositional generalization, and summarizes an approach. The first part contains concepts and properties. The second part looks into a machine learning approach. The approach uses architecture design and regularization to regulate information of representations. This report focuses on basic ideas with intuitive and illustrative explanations. We hope this work would be helpful to clarify fundamentals of compositional generalization and lead to advance artificial intelligence.
background contribution method result interpretation main claim
| # | sentence | role | the authors' own work | states the main conclusion | strength |
|---|---|---|---|---|---|
| s0 | Compositional generalization is the capacity to recognize and imagine a large amount of novel combinations from known components. | background 1.00 | 0.05 | 0.04 | asserted 1.00 |
| s1 | It is a key in human intelligence, but current neural networks generally lack such ability. | background 0.70 | 0.08 | 0.06 | asserted 0.99 |
| s2 | This report connects a series of our work for compositional generalization, and summarizes an approach. | contribution 1.00 | 0.63 | 0.07 | asserted 0.98 |
| s3 | The first part contains concepts and properties. | contribution 0.90 | 0.22 | 0.03 | asserted 0.98 |
| s4 | The second part looks into a machine learning approach. | contribution 0.72 | 0.41 | 0.03 | asserted 0.96 |
| s5 | The approach uses architecture design and regularization to regulate information of representations. | method 0.99 | 0.63 | 0.11 | asserted 1.00 |
| s6 | This report focuses on basic ideas with intuitive and illustrative explanations. | contribution 0.61 | 0.22 | 0.04 | asserted 0.99 |
| s7 | We hope this work would be helpful to clarify fundamentals of compositional generalization and lead to advance artificial intelligence. | future work 0.82 | 0.05 | 0.15 | hedged possibility 0.81 |
What was examined, and what was not
- arXiv listing abstract v1
- full text
- figures and tables
- references and citing sentences
- relations to other papers
- P abstract
- S sections
- T body sentences
- C citations
- R cross-paper relations
The paper
Original abstract
Compositional generalization is the capacity to recognize and imagine a large amount of novel combinations from known components. It is a key in human intelligence, but current neural networks generally lack such ability. This report connects a series of our work for compositional generalization, and summarizes an approach. The first part contains concepts and properties. The second part looks into a machine learning approach. The approach uses architecture design and regularization to regulate information of representations. This report focuses on basic ideas with intuitive and illustrative explanations. We hope this work would be helpful to clarify fundamentals of compositional generalization and lead to advance artificial intelligence.
Human review
How the literature uses it
What this paper does with the arXiv papers it cites
What it cites · 4 sentences without a confident label
Uses the method or tool of 1 paper
-
The second loss is from stochastic sampling [ref] .
Opening the Black Box of Deep Neural Networks via Information · Summary of four losses · 0.83
Cites as background 7 papers
-
Different compositional generalization approaches have been investigated, such as architecture design [ref] , independence assumption [ref] , data augmentation [ref] , causality [ref] , reinforcement learning [ref] , group theory [ref] and meta-learning [ref] .
Learning to Recombine and Resample Data for Compositional Generalization · Introduction · 0.99
-
Different compositional generalization approaches have been investigated, such as architecture design [ref] , independence assumption [ref] , data augmentation [ref] , causality [ref] , reinforcement learning [ref] , group theory [ref] and meta-learning [ref] .
Understanding disentangling in $\beta$-VAE · Introduction · 0.99
-
Different compositional generalization approaches have been investigated, such as architecture design [ref] , independence assumption [ref] , data augmentation [ref] , causality [ref] , reinforcement learning [ref] , group theory [ref] and meta-learning [ref] .
Compositional generalization in a deep seq2seq model by separating syntax and semantics · Introduction · 0.99
Discussion
Know something about this paper? Leave a comment, or highlight any passage above to comment on it.