Abstract:Vision models can achieve strong performance on classification tasks, but the internal representations supporting their predictions are often difficult to interpret. This work investigates whether sparse autoencoders can decompose intermediate representations of a vision model into interpretable features. We train a ConvNeXt classifier on the FGVC-Aircraft dataset, extract spatial activations from its final feature stage, and train a sparse autoencoder on these activations. The learned sparse features are analyzed using top-activating image patches, activation strength, and class selectivity. Qualitative visual inspection reveals that several features correspond to recognizable aircraft structures and visual patterns. We evaluate a subset of selected features using input-space and feature-space ablations, measuring how blurring image patches and suppressing sparse features affect class logits, classification margins, and prediction confidence. The results suggest that sparse autoencoders can reveal partially interpretable, class-relevant visual features associated with aircraft recognition, while also exposing limitations such as polysemanticity and coarse spatial localization.
From: Deepshik Sharma [view email]
[v1]
Sat, 13 Jun 2026 20:54:42 UTC (1,434 KB)
此内容由惯性聚合(RSS阅读器)自动聚合整理，仅供阅读参考。 原文来自 — 版权归原作者所有。