基于机器学习的广告学歧视信息检测模型
网络出版日期: 2026-05-29
基金资助
中央高校基本科研业务费专项资金(3262025T40);省部级网络安全与人工智能研究基地科研经费
版权
Machine learning-based detection model for discriminatory information in advertising
Online published: 2026-05-29
Copyright
随着生成式人工智能技术深度渗透至广告内容创作与投放环节,其在提升效率与个性化水平的同时,也引入了新型且复杂的歧视性内容风险,对广告的公平性、品牌安全及用户体验构成严峻挑战。针对广告图像中潜在歧视性内容识别难的问题,本文提出了一种新型深度卷积神经网络架构模型——ResNet-SPP-ResNet。该模型在结构上融合双重残差网络(ResNet)与空间金字塔池化(SPP)机制,构建对称式网络体系,以增强多尺度特征表达能力。为支撑模型训练与验证,本文构建了广告图像数据集AD-IMAGE-2024,涵盖近5年真实广告样本,并完成歧视标签标注与标准化处理。在该数据集上的对比实验表明,所提模型在二元分类任务中取得了优于主流模型的性能表现,其中准确率为
孔子潇 , 方梦浩 , 张晓恰 , 王亚杰 , 唐湘云 , 徐杨 . 基于机器学习的广告学歧视信息检测模型[J]. 网络空间安全科学学报, 2026 , 4(2) : 90 -99 . DOI: 10.20172/j.issn.2097-3136.260407
With the deep integration of generative artificial intelligence into advertising content creation and delivery, it not only improves efficiency and personalization but also introduces new and complex risks of discriminatory content. To address the difficulty in identifying potential discriminatory content in advertising images, this paper proposes a novel deep convolutional neural network architecture, termed ResNet-SPP-ResNet. The model integrates dual residual networks (ResNet) with a spatial pyramid pooling (SPP) mechanism to form a symmetric network structure, thereby enhancing multi-scale feature representation. To support model training and evaluation, we constructed a new advertising image dataset, AD-IMAGE-2024, which consists of real-world advertisement samples collected over the past five years, with annotated discriminatory labels and standardized preprocessing. Comparative experiments conducted on this dataset demonstrate that the proposed model achieves superior performance in binary classification tasks, with an accuracy of
表 1 消融实验的二元分类结果Table 1 Binary classification results of the ablation test |
| 算法 | 准确率 | 精确率 | 召回率 | F1值 |
| SPP-ResNet | ||||
| SPP | ||||
| ResNet | ||||
| 本文 |
表 2 对比实验的二元分类结果Table 2 Binary classification results of the comparison test |
| 算法 | 准确率 | 精确率 | 召回率 | F1值 |
| Transformer | ||||
| Vision Transformer | ||||
| EfficientNet | ||||
| DenseNet | ||||
| 本文 |
表 3 FACET数据集基于肤色等级的性别分类结果Table 3 Result of gender classification based on skin tone levels in the FACET dataset |
| 算法 | 肤色等级 | 准确率 |
| ResNet | 1 | |
| 2 | ||
| 3 | ||
| 4 | ||
| 5 | ||
| 6 | ||
| 7 | ||
| 8 | ||
| 9 | ||
| Vision Transformer | 1 | |
| 2 | ||
| 3 | ||
| 4 | ||
| 5 | ||
| 6 | ||
| 7 | ||
| 8 | ||
| 9 | ||
| DenseNet | 1 | |
| 2 | ||
| 3 | ||
| 4 | ||
| 5 | ||
| 6 | ||
| 7 | ||
| 8 | ||
| 9 | ||
| ResNet-SPP- ResNet | 1 | |
| 2 | ||
| 3 | ||
| 4 | ||
| 5 | ||
| 6 | ||
| 7 | ||
| 8 | ||
| 9 |
| 1 |
Shah N, Engineer S, Bhagat N, et al. Research trends on the usage of machine learning and artificial intelligence in advertising[J]. Augmented Human Research, 2020, 5, 19.
|
| 2 |
Datta A, Datta A, Makagon J, et al. Discrimination in online advertising: a multidisciplinary inquiry[C]//Conference on Fairness, Accountability and Transparency. PMLR, 2018: 20-34.
|
| 3 |
Dalenberg D J. Preventing discrimination in the automated targeting of job advertisements[J]. Computer Law & Security Review, 2018, 34 (3): 615- 627.
|
| 4 |
Friedmann E, Solodoha E, Loureiro S M C, et al. Disidentification: the long-term effects of offensive-discriminatory advertising[J]. Psychology & Marketing, 2025, 42 (11): 2767- 2788.
|
| 5 |
He K M, Zhang X Y, Ren S Q, et al. Deep residual learning for image recognition[C]//Proceedings of the 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR). Piscataway: IEEE Press, 2016: 770-778.
|
| 6 |
He K M, Zhang X Y, Ren S Q, et al. Spatial pyramid pooling in deep convolutional networks for visual recognition[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2015, 37 (9): 1904- 1916.
|
| 7 |
Github[EB/OL].[2025-12-10]https://github.com/poshangcun13/AD-IMAGE-2024.git.
|
| 8 |
Vaswani A, Shazeer N, Parmar N, et al. Attention is all you need[PP/OL]. V7. arXiv (2023-08-02)[2024-08-25]. https://doi.org/10.48550/arXiv.1706.03762.
|
| 9 |
Dosovitskiy A, Beyer L, Kolesnikov A, et al. An image is worth 16×16 words: transformers for image recognition at scale[PP/OL]. V2. arXiv (2021-06-03)[2024-08-25]. https://doi.org/10.48550/arXiv.2010.11929.
|
| 10 |
Tan M X, Le Q V. EfficientNet: rethinking model scaling for convolutional neural networks[PP/OL]. V5. arXiv (2020-09-11)[2024-08-25]. https://doi.org/10.48550/arXiv.1905.11946.
|
| 11 |
Huang G, Liu Z, Van D M, et al. Densely connected convolutional networks[C]//Proceedings of the 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR). Piscataway: IEEE Press, 2017: 2261-2269.
|
| 12 |
金增旺, 江令洋, 丁俊怡, 等. 工业控制系统安全研究综述[J]. 信息网络安全, 2025, 25 (3): 341- 363.
Jin Z W, Jiang L Y, Ding J Y, et al. A review of research on industrial control system security[J]. Netinfo Security, 2025, 25 (3): 341- 363.
|
| 13 |
陈红松, 刘新蕊, 陶子美, 等. 基于深度学习的时序数据异常检测研究综述[J]. 信息网络安全, 2025, 25 (3): 364- 391.
Chen H S, Liu X R, Tao Z M, et al. A survey of anomaly detection model for time series data based on deep learning[J]. Netinfo Security, 2025, 25 (3): 364- 391.
|
| 14 |
Tariq M, Khan M A. Offensive advertising: a religion based Indian study[J]. Journal of Islamic Marketing, 2017, 8 (4): 656- 668.
|
| 15 |
朱启瑞, 陈荣华, 杨哲慜, 等. 移动社交应用跨用户隐私分享行为的最小必要合规检测方法[J]. 网络空间安全科学学报, 2024, 2 (3): 3- 12.
Zhu Q R, Chen R H, Yang Z M, et al. Inspection method of data minimization compliance for cross-user privacy sharing behavior in mobile social applications[J]. Journal of Cybersecurity, 2024, 2 (3): 3- 12.
|
| 16 |
张芃芃, 宋宗泽, 彭勃, 等. 面向人脸深度伪造检测模型的校准性评测[J]. 网络空间安全科学学报, 2023, 1 (3): 97- 106.
Zhang P P, Song Z Z, Peng B, et al. Calibration evaluation of models oriented to deepfake detection[J]. Journal of Cybersecurity, 2023, 1 (3): 97- 106.
|
| 17 |
Choi J A, Lim K. Identifying machine learning techniques for classification of target advertising[J]. ICT Express, 2020, 6 (3): 175- 180.
|
| 18 |
Shaqoor N A, Kuppusamy K S. Machine learning based heterogeneous web advertisements detection using a diverse feature set[J]. Future Generation Computer Systems, 2018, 89, 68- 77.
|
| 19 |
Huang T H, Yu C M, Kao H Y. Data-driven and deep learning methodology for deceptive advertising and phone scams detection[C]//Proceedings of the 2017 Conference on Technologies and Applications of Artificial Intelligence (TAAI). Piscataway: IEEE Press, 2017: 166-171.
|
| 20 |
Wannakan N, Lertnattee V. Using machine learning for detection of illegal food advertising text[D]. Thailand: Silpakorn University, 2020.
|
| 21 |
Phaisangittisagul E, Koobkrabee Y, Wirojborisuth K, et al. Target advertising classification using combination of deep learning and text model[C]//Proceedings of the 2019 10th International Conference of Information and Communication Technology for Embedded Systems (IC-ICTES). Piscataway: IEEE Press, 2019: 1-4.
|
| 22 |
Frissen R, Adebayo K J, Nanda R. A machine learning approach to recognize bias and discrimination in job advertisements[J]. AI & Society, 2023, 38 (2): 1025- 1038.
|
| 23 |
Li Z, Zhang W Y, Zhang H T, et al. Global digital compact: a mechanism for the governance of online discriminatory and misleading content generation[J]. International Journal of Human–Computer Interaction, 2025, 41 (2): 1381- 1396.
|
| 24 |
Zhao J D, Lei W, Li Z J, et al. Detection of crowdedness in bus compartments based on ResNet algorithm and video images[J]. Multimedia Tools and Applications, 2022, 81 (4): 4753- 4780.
|
| 25 |
Sarwinda D, Paradisa R H, Bustamam A, et al. Deep learning in image classification using residual network (ResNet) variants for detection of colorectal cancer[J]. Procedia Computer Science, 2021, 179, 423- 431.
|
| 26 |
Zheng Z, Zhang H, Li X J, et al. ResNet-based model for cancer detection[C]//Proceedings of the 2021 IEEE International Conference on Consumer Electronics and Computer Engineering (ICCECE). Piscataway: IEEE Press, 2021: 325-328.
|
| 27 |
Gao H, Zhen T, Li Z H. Detection of wheat unsound kernels based on improved ResNet[J]. IEEE Access, 2022, 10, 20092- 20101.
|
| 28 |
邢长友, 王梓澎, 张国敏, 等. 基于预训练Transformers的物联网设备识别方法[J]. 信息网络安全, 2024, 24 (8): 1277- 1290.
Xing C Y, Wang Z P, Zhang G M, et al. IoT device identification method based on pre-trained transformers[J]. Netinfo Security, 2024, 24 (8): 1277- 1290.
|
| 29 |
Gustafson L, Rolland C, Ravi N, et al. FACET: fairness in computer vision evaluation benchmark[C]//Proceedings of the 2023 IEEE/CVF International Conference on Computer Vision (ICCV). Piscataway: IEEE Press, 2023: 20313-20325.
|
/
| 〈 |
|
〉 |