Abstraction Learning

TL;DR

This paper introduces a novel framework called ONE that learns abstraction directly through a network evolution algorithm, aiming to bridge the gap between artificial and human intelligence by capturing key elements of human cognition.

Contribution

It proposes a new method for learning abstraction without human-defined rules, combining a partition structure, constrained optimization, and network evolution.

Findings

Demonstrates elementary human-like intelligence on MNIST

Achieves low energy consumption and lifelong learning

Enables knowledge sharing in neural networks

Abstract

There has been a gap between artificial intelligence and human intelligence. In this paper, we identify three key elements forming human intelligence, and suggest that abstraction learning combines these elements and is thus a way to bridge the gap. Prior researches in artificial intelligence either specify abstraction by human experts, or take abstraction as a qualitative explanation for the model. This paper aims to learn abstraction directly. We tackle three main challenges: representation, objective function, and learning algorithm. Specifically, we propose a partition structure that contains pre-allocated abstraction neurons; we formulate abstraction learning as a constrained optimization problem, which integrates abstraction properties; we develop a network evolution algorithm to solve this problem. This complete framework is named ONE (Optimization via Network Evolution). In our experiments on MNIST, ONE shows elementary human-like intelligence, including low energy consumption, knowledge sharing, and lifelong learning.