Measuring Compositional Generalization: A Comprehensive Method on Realistic Data

12/20/2019
by   Daniel Keysers, et al.
25

State-of-the-art machine learning methods exhibit limited compositional generalization. At the same time, there is a lack of realistic benchmarks that comprehensively measure this ability, which makes it challenging to find and evaluate improvements. We introduce a novel method to systematically construct such benchmarks by maximizing compound divergence while guaranteeing a small atom divergence between train and test sets, and we quantitatively compare this method to other approaches for creating compositional generalization benchmarks. We present a large and realistic natural language question answering dataset that is constructed according to this method, and we use it to analyze the compositional generalization ability of three machine learning architectures. We find that they fail to generalize compositionally and that there is a surprisingly strong negative correlation between compound divergence and accuracy. We also demonstrate how our method can be used to create new compositionality benchmarks on top of the existing SCAN dataset, which confirms these findings.

READ FULL TEXT

page 20

page 34

research
10/15/2020

Hierarchical Poset Decoding for Compositional Generalization in Language

We formalize human language understanding as a structured prediction tas...
research
07/14/2020

A novel dowscaling procedure for compositional data in the Aitchison geometry with application to soil texture data

In this work, we present a novel downscaling procedure for compositional...
research
07/17/2020

Compositional Generalization in Semantic Parsing: Pre-training vs. Specialized Architectures

While mainstream machine learning methods are known to have limited abil...
research
12/15/2020

*-CFQ: Analyzing the Scalability of Machine Learning on a Compositional Task

We present *-CFQ ("star-CFQ"): a suite of large-scale datasets of varyin...
research
04/28/2022

Toward Compositional Generalization in Object-Oriented World Modeling

Compositional generalization is a critical ability in learning and decis...
research
04/07/2022

Compositional Generalization and Decomposition in Neural Program Synthesis

When writing programs, people have the ability to tackle a new complex t...
research
02/19/2019

Measuring Compositionality in Representation Learning

Many machine learning algorithms represent input data with vector embedd...

Please sign up or login with your details

Forgot password? Click here to reset