最新研究揭示LLM中两类微妙偏见——刻板印象与偏离，量化评估方法出炉
arXiv:2508.06649v3 Announce Type: replace Abstract: Large language models (LLMs) are widely applied across diverse domains, raising concerns about the…
深度神经网络公平性修复新突破，提供可证明保证，已被ASE 2025录用。
arXiv:2605.19549v1 Announce Type: cross Abstract: Deep neural networks (DNNs) are suffering from ethical issues such as individual discrimination. In …
提出新方法检测生成式AI中的因果偏差，为提升模型公平性提供理论支撑。
arXiv:2605.11365v2 Announce Type: replace-cross Abstract: Automated systems built on artificial intelligence (AI) are increasingly deployed across hig…
可微分优化层为深度学习中的公平性提供数学保证，ICML 2026前沿研究。
arXiv:2605.17118v1 Announce Type: new Abstract: Differentiable optimization layers are traditionally integrated in predict-then-optimize frameworks wh…
新研究揭示共形预测池化校准隐藏组间异质性，推导守恒律指出公平性代价
arXiv:2605.14260v2 Announce Type: replace-cross Abstract: Conformal prediction is often calibrated with a single pooled threshold, but this can hide c…
突破传统结果公平，提出GESD度量过程解释稳定性，揭示算法偏见新维度
arXiv:2605.15295v1 Announce Type: cross Abstract: Machine learning (ML) algorithms are increasingly deployed in high-stakes decision-making domains su…
LLM输出看似公平但内部仍藏偏见，研究发现不同群体间因果效力不对称，警惕高风决策风险。
arXiv:2605.15217v1 Announce Type: new Abstract: Instruction-tuned language models exhibit behavioural fairness in high-stakes decisions while retainin…