Multimodal learning for image popularity prediction on social media
Multimodal learning for image popularity prediction on social media
复制标题
DOI：
10.1109/icce-tw.2016.7521017
复制
发表时间：
2016-05
期刊：
影响因子：
--
通讯作者：
Jiani Hu;T. Yamasaki;K. Aizawa
中科院分区：
文献类型：
--
作者：
Jiani Hu;T. Yamasaki;K. Aizawa
每天有数十亿张照片通过图片分享服务上传到互联网上，例如Flickr、Instagram等。这些社交媒体的规模不断扩大，给社会人气预测带来了新的挑战。在本文中，我们尝试使用一种多模态学习方法，该方法同时使用标签特征和视觉特征进行流行度预测。我们在实验中比较了几种多模态方法和单模态方法，并给出了结果。
Billions of photographs are uploaded to the Internet every day through image sharing services, e.g., Flickr, Instagram, etc. The growing size of these social media poses new challenges in social popularity prediction. In this paper, we try to use a multimodal learning approach which uses both tag feature and visual feature for popularity prediction. We compare several multimodal approaches with unimodal approaches in our experiments and show the results.