Newest papers

Post-Hoc Feature Selection Layer for Neural Networks Interpretability

João Kenji Suwa, João Pedro Silveira e Silva, Bruno Iochins Grisci · 2026

DOI: 10.1109/ijcnn69854.2026.11729834 · cited by 0 works (OpenAlex) · open-access PDF

Arabic

إن التفكير في الشبكات العصبية المعقدة لا يزال مشكلًا، خصوصًا بالنسبة إلى النماذج التي تم تطويرها بالفعل في مجالات عالية الدخل. لمناقشة هذا، قمنا بتطوير طبقة اختيار الخصائص (FSL) لتعزيز وزن الخصائص بعد التدريب، وهو نوع من التفكير الذي نسميه FSL بعد التدريب. يبدو أن الطريقة التي نقوم بها هي التفكير في FSL كوسيلة سهلة ومتعددة الخصائص التي تم تطويرها في مجالات عالية الدخل، والتي تتكامل مع النماذج التي تم تطويرها بالفعل قبل التدريب على مجموعات البيانات الأصلية لتعزيز الخصائص التي يعتبرها النماذج الأصلية الأكثر أهمية. لقد أدركت هذه FSL بعد التد

Machine translation (Meta M2M100); the English original follows.

Abstract

The interpretability of complex neural networks remains a critical challenge, especially for models already deployed in high-stakes domains. To address this, we adapted the Feature Selection Layer (FSL) to optimize feature weights post-training, a variant we term post-hoc FSL. Our approach reframes the FSL as a lightweight, trainable module that integrates with already frozen pre-trained models on tabular datasets to highlight the features the original model considers most important. This post-hoc FSL learns feature relevance by fine-tuning its weights based on the original model's learned outputs. Crucially, this process is non-invasive, operating without altering the original model's architecture or its learned parameters. We conducted our experiments using both statistical and visual metrics, including accuracy, F1 score, recall, precision, weighted t-SNE and silhouette score, and also analyzed the stability of the post-hoc FSL on high-dimensional synthetic and real-world tabular datasets. We compared the post-hoc FSL feature weighting method using these metrics against the original embedded FSL and other post-hoc interpretability methods, such as Integrated Gradients, Noise Tunnel, DeepLIFT, Gradient SHAP, and Feature Ablation. Experimental results demonstrate that the post-hoc FSL feature weighting method successfully identified relevant features across the different datasets, maintaining the predictive power of the original neural network while enhancing its interpretability.

Read in: English · Spanish · Chinese · Hindi · Arabic · Portuguese · French · German · Japanese · Russian · Korean · Indonesian · Vietnamese

Topics: Interpretability, Artificial intelligence, Computer science, Artificial neural network, Feature selection, Layer (electronics)

Sources: OpenAlex (CC0), arXiv, Crossref. Agent Exchange