本研究针对AI模型评估中因数据捷径（shortcut learning）导致的性能误判问题，提出捷径壳学习（SHL）诊断范式与无捷径评估框架（SFEF）。通过构建拓扑数据集验证发现，传统认为全局能力较弱的卷积模型（CNN）实际优于Transformer模型，挑战了现有认知，为AI可靠性 ...

If your Uber driver takes a shortcut, you might get to your destination faster. But if a machine learning model takes a shortcut, it might fail in unexpected ways. In machine learning, a shortcut ...

为解决中文自然语言处理(NLP)模型中因数据偏差和捷径学习(shortcut learning)导致的分布外泛化能力不足问题，兰州科研团队提出CAT-CDA框架。该研究通过梯度显著性分析定位因果特征，结合非似然对抗训练(Non-likelihood Adversarial Training)自动生成标签反转的反事实数据。

If your Uber driver takes a shortcut, you might get to your destination faster. But if a machine learning model takes a shortcut, it might fail in unexpected ways. In machine learning, a shortcut ...

CAMBRIDGE, Mass. — If your Uber driver takes a shortcut, you might get to your destination faster. But if a machine learning model takes a shortcut, it might fail in unexpected ways. In machine ...

Machine-learning models highlight text markers on radiographs as important for COVID-19 prediction. Swapping these markers between a pair of COVID-19-positive and COVID-19-negative images ...

In the age of AI tutors, the way we study is undergoing a quiet but radical transformation. Just a year ago, students were asking ChatGPT to explain tough homework questions or summarize dense ...