Recommending Themes for Ad Creative Design via Visual-Linguistic Representations

被引:7
|
作者
Zhou, Yichao [1 ]
Mishra, Shaunak [2 ]
Verma, Manisha [2 ]
Bhamidipati, Narayan [2 ]
Wang, Wei [1 ]
机构
[1] Univ Calif Los Angeles, Los Angeles, CA 90024 USA
[2] Yahoo Res, Sunnyvale, CA USA
关键词
Online advertising; transformers; visual-linguistic representation;
D O I
10.1145/3366423.3380001
中图分类号
TP [自动化技术、计算机技术];
学科分类号
0812 ;
摘要
There is a perennial need in the online advertising industry to refresh ad creatives, i.e., images and text used for enticing online users towards a brand. Such refreshes are required to reduce the likelihood of ad fatigue among online users, and to incorporate insights from other successful campaigns in related product categories. Given a brand, to come up with themes for a new ad is a painstaking and time consuming process for creative strategists. Strategists typically draw inspiration from the images and text used for past ad campaigns, as well as world knowledge on the brands. To automatically infer ad themes via such multimodal sources of information in past ad campaigns, we propose a theme (keyphrase) recommender system for ad creative strategists. The theme recommender is based on aggregating results from a visual question answering (VQA) task, which ingests the following: (i) ad images, (ii) text associated with the ads as well as Wikipedia pages on the brands in the ads, and (iii) questions around the ad. We leverage transformer based cross-modality encoders to train visual-linguistic representations for our VQA task. We study two formulations for the VQA task along the lines of classification and ranking; via experiments on a public dataset, we show that cross-modal representations lead to significantly better classification accuracy and ranking precision-recall metrics. Cross-modal representations show better performance compared to separate image and text representations. In addition, the use of multimodal information shows a significant lift over using only textual or visual information.
引用
收藏
页码:2521 / 2527
页数:7
相关论文
共 6 条
  • [1] Visual Relationship Detection With Visual-Linguistic Knowledge From Multimodal Representations
    Chiou, Meng-Jiun
    Zimmermann, Roger
    Feng, Jiashi
    [J]. IEEE ACCESS, 2021, 9 : 50441 - 50451
  • [2] Compressing Visual-linguistic Model via Knowledge Distillation
    Fang, Zhiyuan
    Wang, Jianfeng
    Hu, Xiaowei
    Wang, Lijuan
    Yang, Yezhou
    Liu, Zicheng
    [J]. 2021 IEEE/CVF INTERNATIONAL CONFERENCE ON COMPUTER VISION (ICCV 2021), 2021, : 1408 - 1418
  • [3] CLIP-Gaze: Towards General Gaze Estimation via Visual-Linguistic Model
    Yin, Pengwei
    Zeng, Guanzhong
    Wang, Jingjing
    Xie, Di
    [J]. THIRTY-EIGHTH AAAI CONFERENCE ON ARTIFICIAL INTELLIGENCE, VOL 38 NO 7, 2024, : 6729 - 6737
  • [4] Automated Construction of Visual-Linguistic Knowledge via Concept Learning from Cartoon Videos
    Ha, Jung-Woo
    Kim, Kyung-Min
    Zhang, Byoung-Tak
    [J]. PROCEEDINGS OF THE TWENTY-NINTH AAAI CONFERENCE ON ARTIFICIAL INTELLIGENCE, 2015, : 522 - 528
  • [5] Faster Zero-shot Multi-modal Entity Linking via Visual-Linguistic Representation
    Qiushuo Zheng
    Hao Wen
    Meng Wang
    Guilin Qi
    Chaoyu Bai
    [J]. Data Intelligence, 2022, 4 (03) : 493 - 508
  • [6] Faster Zero-shot Multi-modal Entity Linking via Visual-Linguistic Representation
    Zheng, Qiushuo
    Wen, Hao
    Wang, Meng
    Qi, Guilin
    Bai, Chaoyu
    [J]. DATA INTELLIGENCE, 2022, 4 (03) : 493 - 508