When a Zero-Shooter Cheats: Improving Age Estimation via Activation Steering

• URL: http://arxiv.org/abs/2605.17658v1
• Date: Sun, 17 May 2026 21:20:36 GMT
• Title: When a Zero-Shooter Cheats: Improving Age Estimation via Activation Steering
• Abstract summary: We find that vision-language models (VLMs) tend to identify the depicted person and infer their age from memorized knowledge.<n>This phenomenon leads to substantially incorrect predictions when non-celebrities are misidentified as celebrities.<n>We propose an activation steering method that suppresses the shortcut by intervening on the hidden states of the VLM.
• Score: 4.931984026341269
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: Different age-related regulations have been proposed to protect minors from harmful content and interactions online. Automated age estimation is central to enforcing such regulations, and vision-language models (VLMs) achieve state-of-the-art performance on this task. However, we find that the zero-shot nature of VLM-based age estimation produces an unexpected side effect we call the identity shortcut: Instead of estimating age from visual features, VLMs tend to identify the depicted person and infer their age from memorized knowledge. This phenomenon leads to substantially incorrect predictions when non-celebrities are misidentified as celebrities. It also produces deceptively high robustness to noise and adversarial perturbations on celebrity images, which dominate popular benchmarks. To mitigate this, we propose an activation steering method that suppresses the shortcut by intervening on the hidden states of the VLM. This method improves age estimation accuracy for both memorized and unseen identities, reducing mean absolute error by up to 25% across popular benchmarks.

Related papers

• MeInTime: Bridging Age Gap in Identity-Preserving Face Restoration [24.354251784441217]
  MeInTime is a diffusion-based face restoration method that extends reference-based restoration from same-age to cross-age settings.<n>Given one or few reference images along with an age prompt corresponding to the degraded input, MeInTime achieves faithful restoration with both identity fidelity and age consistency.
  arXiv  Detail & Related papers  (2026-03-19T09:11:07Z)
• The Illusion of Forgetting: Attack Unlearned Diffusion via Initial Latent Variable Optimization [51.835894707552946]
  Unlearning-based defenses claim to purge Not-Safe-For-Work concepts from diffusion models (DMs)<n>We show that unlearning partially disrupts the mapping between linguistic symbols and the underlying knowledge, which remains intact as dormant memories.<n>We propose IVO, a concise and powerful attack framework that reactivates these dormant memories by reconstructing the broken mappings.
  arXiv  Detail & Related papers  (2026-01-30T02:39:51Z)
• Counterfactual VLA: Self-Reflective Vision-Language-Action Model with Adaptive Reasoning [71.19675094463834]
  This work introduces Counterfactual VLA, a self-reflective VLA framework that enables the model to reason about and revise its planned actions before execution.<n>CF-VLA first generates time-segmented meta-actions that summarize driving intent, and then performs counterfactual reasoning conditioned on both the meta-actions and the visual context.<n> Experiments on large-scale driving datasets show that CF-VLA improves trajectory accuracy by up to 17.6%, enhances safety metrics by 20.5%, and exhibits adaptive thinking.
  arXiv  Detail & Related papers  (2025-12-30T19:04:17Z)
• Towards an Automated Framework to Audit Youth Safety on TikTok [6.83645418303131]
  This paper investigates the effectiveness of TikTok's enforcement mechanisms for limiting the exposure of harmful content to youth accounts.<n>We collect over 7000 videos, classify them as harmful vs not-harmful, and then simulate interactions using age-specific sockpuppet accounts.<n>Preliminary results show minimal differences in content exposure between adult and youth accounts, raising concerns about the platform's age-based moderation.
  arXiv  Detail & Related papers  (2025-09-06T21:30:11Z)
• CILF-CIAE: CLIP-driven Image-Language Fusion for Correcting Inverse Age Estimation [14.639340916340801]
  The age estimation task aims to predict the age of an individual by analyzing facial features in an image. Existing CLIP-based age estimation methods require high memory usage and lack an error feedback mechanism. We propose a novel CLIP-driven Image-Language Fusion for Correcting Inverse Age Estimation (CILF-CIAE)
  arXiv  Detail & Related papers  (2023-12-04T09:35:36Z)
• Adaptive Mean-Residue Loss for Robust Facial Age Estimation [7.667560350473354]
  We propose a loss function for robust facial age estimation via distribution learning. Experimental results in the datasets FG-NET and CLAP2016 have validated the effectiveness of the proposed loss.
  arXiv  Detail & Related papers  (2022-03-31T16:28:34Z)
• LAE : Long-tailed Age Estimation [52.5745217752147]
  We first formulate a simple standard baseline and build a much strong one by collecting the tricks in pre-training, data augmentation, model architecture, and so on. Compared with the standard baseline, the proposed one significantly decreases the estimation errors. We propose a two-stage training method named Long-tailed Age Estimation (LAE), which decouples the learning procedure into representation learning and classification.
  arXiv  Detail & Related papers  (2021-10-25T09:05:44Z)
• FP-Age: Leveraging Face Parsing Attention for Facial Age Estimation in the Wild [50.8865921538953]
  We propose a method to explicitly incorporate facial semantics into age estimation. We design a face parsing-based network to learn semantic information at different scales. We show that our method consistently outperforms all existing age estimation methods.
  arXiv  Detail & Related papers  (2021-06-21T14:31:32Z)
• Age Gap Reducer-GAN for Recognizing Age-Separated Faces [72.26969872180841]
  We propose a novel algorithm for matching faces with temporal variations caused due to age progression. The proposed generative adversarial network algorithm is a unified framework that combines facial age estimation and age-separated face verification.
  arXiv  Detail & Related papers  (2020-11-11T16:43:32Z)
• Temporal Sparse Adversarial Attack on Sequence-based Gait Recognition [56.844587127848854]
  We demonstrate that the state-of-the-art gait recognition model is vulnerable to such attacks. We employ a generative adversarial network based architecture to semantically generate adversarial high-quality gait silhouettes or video frames. The experimental results show that if only one-fortieth of the frames are attacked, the accuracy of the target model drops dramatically.
  arXiv  Detail & Related papers  (2020-02-22T10:08:42Z)

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.