让神经元自带概率模型，能捕捉不确定性。
uGMM-NN: Univariate Gaussian Mixture Model Neural Network
- 每个神经元用可学习的高斯混合模型表示激活值
- 在分类任务上性能接近传统网络，还能输出概率解释
- 适合需要可信度评估的场景，如医疗与自动驾驶
本文提出一种新型神经网络架构——单变量高斯混合模型神经网络（uGMM-NN），将概率推理直接嵌入深度网络的计算单元中。与传统神经元先进行加权求和再应用固定非线性不同，uGMM-NN 的每个节点将激活值建模为一个单变量高斯混合分布，包含可学习的均值、方差和混合系数。该设计使神经元能捕捉多模态特征与不确定性，同时保持前馈网络的可扩展性。实验表明，uGMM-NN 在分类任务上表现与传统多层感知机相当，且能提供激活值的概率解释。该框架为现代神经网络引入不确定性感知组件提供了基础，拓展了判别与生成建模的新方向。
原文摘要 · Abstract (English)
This paper introduces the Univariate Gaussian Mixture Model Neural Network (uGMM-NN), a novel neural architecture that embeds probabilistic reasoning directly into the computational units of deep networks. Unlike traditional neurons, which apply weighted sums followed by fixed non-linearities, each uGMM-NN node parameterizes its activations as a univariate Gaussian mixture, with learnable means, variances, and mixing coefficients. This design enables richer representations by capturing multimodality and uncertainty at the level of individual neurons, while retaining the scalability of standard feed-forward networks. We demonstrate that uGMM-NN can achieve competitive discriminative performance compared to conventional multilayer perceptrons, while additionally offering a probabilistic interpretation of activations. The proposed framework provides a foundation for integrating uncertainty-aware components into modern neural architectures, opening new directions for both discriminative and generative modeling.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。