What type of study is this?

This is a Quantitative Study study.

October 20, 2025Open Access

Task-Agnostic Experts Composition for Continual Learning

LQLuigi QuarantielloUniversity of Pisa ACAndrea CossuUniversity of Pisa VLVincenzo LomonacoUniversity of Pisa

Key Points

Our Expert Composition method achieves significantly higher accuracy compared to baseline algorithms, improving learning outcomes.
The approach utilizes a zero-shot ensemble of expert models to efficiently tackle complex problems with compositionality.
A challenging benchmark evaluates the compositional capabilities of the method, underscoring its effectiveness.
This methodology reduces computational resource needs, suggesting a more sustainable AI framework for continual learning.

Abstract

Compositionality is one of the fundamental abilities of the human reasoning process, that allows to decompose a complex problem into simpler elements. Such property is crucial also for neural networks, especially when aiming for a more efficient and sustainable AI framework. We propose a compositional approach by ensembling zero-shot a set of expert models, assessing our methodology using a challenging benchmark, designed to test compositionality capabilities. We show that our Expert Composition method is able to achieve a much higher accuracy than baseline algorithms while requiring less computational resources, hence being more efficient.

Ask AI

Helpful

Bookmark

View Full Paper