{"ID":2888875,"CreatedAt":"2026-06-01T04:54:23.091178241Z","UpdatedAt":"2026-06-01T04:54:23.091178241Z","DeletedAt":null,"paper_url":"https://arxiv.org/abs/2507.22832","arxiv_id":"2507.22832","title":"Pulling Back the Curtain on Deep Networks","abstract":"In linear models, visualizing a weight vector naturally reveals the model's preferred input direction, but extending this intuition to deep networks via gradients or gradient ascent often yields brittle or adversarial-looking features. We argue that deep networks are better understood as input-conditioned affine operators, whose natural adjoint action pulls a neuron's preferred direction back to input space. We further refine this representation by backward-only softening and iterative enhancement to reconstruct coherent local structures encoded by the target neuron. This provides a unifying perspective on previously disparate ideas such as SmoothGrad, B-cos-style alignment, and Feature Accentuation. The resulting Semantic Pullbacks (SP) generate perceptually aligned, class-conditional post-hoc explanations that emphasize semantically meaningful features, facilitate coherent counterfactual perturbations, and remain theoretically grounded. Across convolutional architectures (ResNet50, VGG) and transformer-based models (PVT), Semantic Pullbacks achieve the best overall trade-off across faithfulness, stability, and target-sensitivity benchmarks, while remaining general, computationally efficient, and readily integrable into existing deep learning pipelines.","short_abstract":"In linear models, visualizing a weight vector naturally reveals the model's preferred input direction, but extending this intuition to deep networks via gradients or gradient ascent often yields brittle or adversarial-looking features. We argue that deep networks are better understood as input-conditioned affine operat...","url_abs":"https://arxiv.org/abs/2507.22832","url_pdf":"https://arxiv.org/pdf/2507.22832v6","authors":"[\"Maciej Satkiewicz\",\"Roberto Corizzo\",\"Marcin Pietroń\"]","published":"2025-07-30T16:47:42Z","proceeding":"cs.LG","tasks":"[\"cs.LG\",\"cs.CV\",\"cs.NE\"]","methods":"[\"Transformer\"]","has_code":false}