This paper presents the "Uncertainty-aware Mixture of Experts" (uMoE), a novel approach designed to address aleatoric uncertainty in the training of predictive models based on Neural Networks (NNs). While existing methods primarily focus on managing uncertainty during infer-ence, uMoE integrates uncertainty directly into the train-ing process. The uMoE approach adopts a "Divide and Conquer" paradigm to partition the uncertain input space into more manageable subspaces. It consists of Expert components, each trained solely on the portion of input uncertainty corresponding to their subspace. On top of the Experts, a Gating Unit, guided by additional infor-mation about the distribution of uncertain inputs across these subspaces, learns to weight the Experts to minimize deviations from the ground truth. Our results highlight that uMoE significantly outperforms baseline methods in handling data uncertainty. Furthermore, we conducted a robustness analysis, illustrating its capability to adapt to varying levels of uncertainty and suggesting optimal threshold parameters. This innovative approach holds wide applicability across diverse data-driven domains, in-cluding biomedical signal processing, autonomous driv-ing, and production quality control.

翻译：本文提出了一种名为“不确定性感知专家混合”（uMoE）的新方法，旨在解决基于神经网络（NNs）的预测模型训练中的偶然不确定性。现有方法主要关注推理过程中的不确定性管理，而uMoE将不确定性直接集成到训练过程中。uMoE采用“分而治之”范式，将不确定的输入空间划分为更易处理的子空间。它由多个专家组件组成，每个组件仅针对其对应子空间的输入不确定性部分进行训练。在专家之上，一个门控单元通过额外关于不确定输入在各子空间中分布的信息进行引导，学习对专家进行加权，以最小化与真实值的偏差。我们的结果表明，uMoE在处理数据不确定性方面显著优于基线方法。此外，我们还进行了鲁棒性分析，展示了其适应不同不确定性水平的能力，并提出了最优阈值参数。这一创新方法在包括生物医学信号处理、自动驾驶和生产质量控制在内的多种数据驱动领域具有广泛适用性。