Concept-Guided Fine-Tuning: Steering ViTs away from Spurious Correlations to Improve Robustness

• URL: http://arxiv.org/abs/2603.08309v2
• Date: Sun, 15 Mar 2026 05:03:19 GMT
• Title: Concept-Guided Fine-Tuning: Steering ViTs away from Spurious Correlations to Improve Robustness
• Abstract summary: Vision Transformers (ViTs) often degrade under distribution shifts because they rely on spurious correlations.<n>We introduce a novel finetuning framework that steers model reasoning toward concept-level semantics.<n>Our method improves robustness across multiple ViT-based models.
• Score: 16.541207762213272
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Vision Transformers (ViTs) often degrade under distribution shifts because they rely on spurious correlations, such as background cues, rather than semantically meaningful features. Existing regularization methods, typically relying on simple foreground-background masks, which fail to capture the fine-grained semantic concepts that define an object (e.g., ``long beak'' and ``wings'' for a ``bird''). As a result, these methods provide limited robustness to distribution shifts. To address this limitation, we introduce a novel finetuning framework that steers model reasoning toward concept-level semantics. Our approach optimizes the model's internal relevance maps to align with spatially grounded concept masks. These masks are generated automatically, without manual annotation: class-relevant concepts are first proposed using an LLM-based, label-free method, and then segmented using a VLM. The finetuning objective aligns relevance with these concept regions while simultaneously suppressing focus on spurious background areas. Notably, this process requires only a minimal set of images and uses half of the dataset classes. Extensive experiments on five out-of-distribution benchmarks demonstrate that our method improves robustness across multiple ViT-based models. Furthermore, we show that the resulting relevance maps exhibit stronger alignment with semantic object parts, offering a scalable path toward more robust and interpretable vision models. Finally, we confirm that concept-guided masks provide more effective supervision for model robustness than conventional segmentation maps, supporting our central hypothesis.

Related papers

• S$^2$COPE: Self-Supervised Concept Discovery via Preference Learning [15.951097715837799]
  We introduce Self-Supervised Concept discOvery via Preference lEarning (model), a label-free framework that resolves this dilemma.<n>Our work suggest that interpretability can emerge through a model's autonomous interaction with incidental visual structures, without any human supervision.
  arXiv  Detail & Related papers  (2026-06-12T16:02:44Z)
• Concept-based Visual Counterfactual Explanations with Diffusion Models [6.539386622516073]
  C-VCE builds the classifier directly into the generative model via a concept bottleneck layer.<n>Our model lets users toggle on/off semantic concepts during sampling, then adjusts minimally relevant image regions.<n>On benchmarks such as CelebA, C-VCE matches or improves flip rates while producing counterfactuals that are visually closer to the input.
  arXiv  Detail & Related papers  (2026-05-05T15:32:50Z)
• ModuSeg: Decoupling Object Discovery and Semantic Retrieval for Training-Free Weakly Supervised Segmentation [8.671473711194741]
  Weakly supervised semantic segmentation aims to achieve pixel-level predictions using image-level labels.<n>We present ModuSeg, a training-free weakly supervised semantic segmentation framework centered on explicitly decoupling object discovery and semantic assignment.
  arXiv  Detail & Related papers  (2026-04-08T12:38:07Z)
• Perception-Aware Multimodal Spatial Reasoning from Monocular Images [57.42071289037214]
  spatial reasoning from monocular images is essential for autonomous driving.<n>Current Vision-Language Models (VLMs) still struggle with fine-grained geometric perception.<n>We propose a perception-aware multimodal reasoning framework that equipsVLMs with explicit object-centric grounding ability.
  arXiv  Detail & Related papers  (2026-03-07T02:05:12Z)
• SGHA-Attack: Semantic-Guided Hierarchical Alignment for Transferable Targeted Attacks on Vision-Language Models [73.19044613922911]
  Large vision-language models (VLMs) are vulnerable to transfer-based adversarial perturbations.<n>We propose SGHA-Attack, a framework that adopts multiple target references and enforces intermediate-layer consistency.<n>Experiments on open-source and commercial black-box VLMs show that SGHA-Attack achieves stronger targeted transferability than prior methods.
  arXiv  Detail & Related papers  (2026-02-02T03:10:41Z)
• Cross-Layer Discrete Concept Discovery for Interpreting Language Models [13.842670153893977]
  Cross-layer VQ-VAE is a framework that uses vector quantization to map representations across layers.<n>Our approach uniquely combines top-k temperature-based sampling during quantization with EMA codebook updates.
  arXiv  Detail & Related papers  (2025-06-24T22:43:36Z)
• MapBERT: Bitwise Masked Modeling for Real-Time Semantic Mapping Generation [15.116320098263149]
  MapBERT is a novel framework designed to model the distribution of unseen spaces.<n>We show that MapBERT achieves state-of-the-art semantic map generation.<n> Experiments on Gibson benchmarks show that MapBERT achieves state-of-the-art semantic map generation.
  arXiv  Detail & Related papers  (2025-06-09T01:55:55Z)
• Interpretable Few-Shot Image Classification via Prototypical Concept-Guided Mixture of LoRA Experts [79.18608192761512]
  Self-Explainable Models (SEMs) rely on Prototypical Concept Learning (PCL) to enable their visual recognition processes more interpretable.<n>We propose a Few-Shot Prototypical Concept Classification framework that mitigates two key challenges under low-data regimes: parametric imbalance and representation misalignment.<n>Our approach consistently outperforms existing SEMs by a notable margin, with 4.2%-8.7% relative gains in 5-way 5-shot classification.
  arXiv  Detail & Related papers  (2025-06-05T06:39:43Z)
• Semantic Image Synthesis via Diffusion Models [174.24523061460704]
  Denoising Diffusion Probabilistic Models (DDPMs) have achieved remarkable success in various image generation tasks.<n>Recent work on semantic image synthesis mainly follows the de facto GAN-based approaches.<n>We propose a novel framework based on DDPM for semantic image synthesis.
  arXiv  Detail & Related papers  (2022-06-30T18:31:51Z)
• Optimizing Relevance Maps of Vision Transformers Improves Robustness [91.61353418331244]
  It has been observed that visual classification models often rely mostly on the image background, neglecting the foreground, which hurts their robustness to distribution changes. We propose to monitor the model's relevancy signal and manipulate it such that the model is focused on the foreground object. This is done as a finetuning step, involving relatively few samples consisting of pairs of images and their associated foreground masks.
  arXiv  Detail & Related papers  (2022-06-02T17:24:48Z)
• Deep Semantic Matching with Foreground Detection and Cycle-Consistency [103.22976097225457]
  We address weakly supervised semantic matching based on a deep network. We explicitly estimate the foreground regions to suppress the effect of background clutter. We develop cycle-consistent losses to enforce the predicted transformations across multiple images to be geometrically plausible and consistent.
  arXiv  Detail & Related papers  (2020-03-31T22:38:09Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.