Neural additive models (NAMs) can improve the interpretability of deep neural networks by handling input features in separate additive sub-networks. However, they lack inherent mechanisms that provide calibrated uncertainties and enable selection of relevant features and interactions. Approaching NAMs from a Bayesian perspective, we enhance them in three primary ways, namely by a) providing credible intervals for the individual additive sub-networks; b) estimating the marginal likelihood to perform an implicit selection of features via an empirical Bayes procedure; and c) enabling a ranking of feature pairs as candidates for second-order interaction in fine-tuned models. In particular, we develop Laplace-approximated NAMs (LA-NAMs), which show improved empirical performance on tabular datasets and challenging real-world medical tasks.

翻译：神经加性模型（NAMs）通过将输入特征分配到独立的加性子网络中处理，能够提升深度神经网络的可解释性。然而，这些模型缺乏内在机制来提供校准后的不确定性估计，也无法自动选择相关特征与交互作用。我们从贝叶斯视角出发，通过三种主要方式增强神经加性模型，具体包括：a）为各独立加性子网络提供置信区间；b）通过经验贝叶斯过程估计边际似然以实现隐式特征选择；c）对特征对进行排序，将其作为微调模型中二阶交互作用的候选。我们特别开发了拉普拉斯近似神经加性模型（LA-NAMs），该模型在表格数据集及具有挑战性的真实世界医学任务中展现出更优的实证性能。