Bongard-LOGO: A New Benchmark for Human-Level Concept Learning and Reasoning

TL;DR

Bongard-LOGO is a new benchmark designed to evaluate human-level concept learning and reasoning in visual cognition, highlighting gaps in current AI models and emphasizing core properties like context-dependence and analogy-making.

Contribution

The paper introduces Bongard-LOGO, a large, human-interpretable visual cognition benchmark generated with a program-guided technique, capturing key properties of human cognition for AI evaluation.

Findings

Deep learning models perform worse than humans on the benchmark.

The benchmark captures core properties of human cognition such as context-dependence and analogy-making.

Current AI models struggle to replicate human-like concept learning and reasoning.

Abstract

Humans have an inherent ability to learn novel concepts from only a few samples and generalize these concepts to different situations. Even though today's machine learning models excel with a plethora of training data on standard recognition tasks, a considerable gap exists between machine-level pattern recognition and human-level concept learning. To narrow this gap, the Bongard problems (BPs) were introduced as an inspirational challenge for visual cognition in intelligent systems. Despite new advances in representation learning and learning to learn, BPs remain a daunting challenge for modern AI. Inspired by the original one hundred BPs, we propose a new benchmark Bongard-LOGO for human-level concept learning and reasoning. We develop a program-guided generation technique to produce a large set of human-interpretable visual cognition problems in action-oriented LOGO language. Our benchmark captures three core properties of human cognition: 1) context-dependent perception, in which the same object may have disparate interpretations given different contexts; 2) analogy-making perception, in which some meaningful concepts are traded off for other meaningful concepts; and 3) perception with a few samples but infinite vocabulary. In experiments, we show that the state-of-the-art deep learning methods perform substantially worse than human subjects, implying that they fail to capture core human cognition properties. Finally, we discuss research directions towards a general architecture for visual reasoning to tackle this benchmark.