A Neuro-Symbolic Benchmark Suite for Concept Quality and Reasoning Shortcuts
This addresses the issue of trustworthiness and generalization in AI models for researchers, though it is incremental as it focuses on benchmarking rather than solving the problem.
The paper tackles the problem of reasoning shortcuts in tasks requiring both learning and reasoning by introducing rsbench, a benchmark suite that provides customizable tasks and metrics, revealing that achieving high-quality concepts in neural and neuro-symbolic models remains unsolved.
The advent of powerful neural classifiers has increased interest in problems that require both learning and reasoning. These problems are critical for understanding important properties of models, such as trustworthiness, generalization, interpretability, and compliance to safety and structural constraints. However, recent research observed that tasks requiring both learning and reasoning on background knowledge often suffer from reasoning shortcuts (RSs): predictors can solve the downstream reasoning task without associating the correct concepts to the high-dimensional data. To address this issue, we introduce rsbench, a comprehensive benchmark suite designed to systematically evaluate the impact of RSs on models by providing easy access to highly customizable tasks affected by RSs. Furthermore, rsbench implements common metrics for evaluating concept quality and introduces novel formal verification procedures for assessing the presence of RSs in learning tasks. Using rsbench, we highlight that obtaining high quality concepts in both purely neural and neuro-symbolic models is a far-from-solved problem. rsbench is available at: https://unitn-sml.github.io/rsbench.