With the popularity of multimedia applications and social networks, various multimedia data (i.e., texts, images, and videos) on the internet have shown exponential growth [1, 5, 6]. By regarding the storage cost and the computation efficiency, it is becoming more and more popular for data owners to employ cloud services [2, 3]. However, the data owners afraid of cloud services to reveal their private information, such as location and financial status. Moreover, the data analysis (such as feature extraction, retrieval, model construction, etc.) may easily leak important private information [4, 7]. For example, recent study in machine learning have demonstrated that sensitive data can be recovered from models. In this case, both cybersecurity and knowledge discovery are extremely important for analyzing big data. This special issue collects some recent studies on current machine learning techniques as well as privacy-preserving data analysis.

中文翻译：

---

隐私保护多媒体数据分析

随着多媒体应用和社交网络的普及，各种多媒体数据（即互联网上的文本、图像和视频）呈指数级增长 [1, 5, 6]。考虑到存储成本和计算效率，数据所有者越来越喜欢使用云服务 [2, 3]。然而，数据所有者害怕云服务泄露他们的私人信息，例如位置和财务状况。此外，数据分析（如特征提取、检索、模型构建等）很容易泄露重要的私人信息[4, 7]。例如，最近的机器学习研究表明，可以从模型中恢复敏感数据。在这种情况下，网络安全和知识发现对于分析大数据都极为重要。本期特刊收集了一些关于当前机器学习技术以及隐私保护数据分析的最新研究。