arXiv cs.AI· Mahmoud Sanad, Mohammed Bakr Sikal, Osama Abboud, Heba Khdr, J\"org Henkel·· 3 小时前AI 评分36
LEX：面向 OOD 韧性运行时预测的 XAI 置信度估计框架
XAI-Driven Confidence Estimation for OOD-Resilient Runtime Prediction
AI 导读
研究者提出 LEX，首个基于事后 XAI 的 ML 资源管理输入级置信度估计框架，用高斯混合模型聚类训练输入嵌入并为每簇绑定预测误差界，无需重训练即可给出细粒度置信信号。在 Intel Alder Lake 上的 LSTM IPS 预测模型和 34 个基准应用评测中，LEX 对未见负载达到约 94% 包含率，置信界比全局基线紧约 17%，且运行时开销可忽略。
正文 · 原文
Abstract:Machine learning-based resource management (RM) models degrade under out-of-distribution (OOD) inputs at runtime, yet existing approaches in the literature rely on a single global error margin that is overly conservative and fails to distinguish reliablepredictions from catastrophic failures. We propose Lightweight Runtime EXplainability (LEX), the first post-hoc XAI framework for input-specific runtime confidence estimation in ML-based RM. LEX clusters the embeddings of the training input space using a Gaussian Mixture Model and associates per-cluster prediction error bounds to input samples, enabling fine-grained runtime confidence signals without retraining. Evaluated on a state-of-the-art LSTM-based IPS forecasting model on Intel Alder Lake, across 34 benchmark applications, LEX achieves ~94% containment on unseen workloads and produces ~17% tighter per-input confidence bounds than global baselines. With its negligible runtime overhead, LEX advances the field by providing explainable and reliable input-specific confidence signals of ML predictions, thereby making ML-based resource management resilient to OOD at runtime.
| Subjects: | Machine Learning (cs.LG); Artificial Intelligence (cs.AI) |
| Cite as: | arXiv:2611.13199 [cs.LG] |
| (or arXiv:2611.13199v1 [cs.LG] for this version) | |
| https://doi.org/10.48550/arXiv.2611.13199 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Mahmoud Sanad [view email]
[v1]
Mon, 9 Nov 2026 20:23:12 UTC (3,391 KB)
来源：arXiv cs.AI · arxiv.org