It also handles new media data types, such as informatics data for digital healthcare. The goal is to build an intelligent media computing platform that provides people with services and empowers them with knowledge.

Areas of Focus:

Multimedia Search and Mining

The explosion of multimedia data brings us new challenges with regards to multimedia content understanding, organization, presentation, and search - especially in terms of scalability. However, both computational costs and performance are still far from satisfactory. This increasing amount of data, along with the massive amount of grassroots users, also provides us with new opportunities to attack these challenges, as well as conventional problems in multimedia analysis and computer vision. This research direction will use all this information, as well as technologies in multimedia content analysis, machine learning, data mining and human-computer interface, to facilitate multimedia search on and across desktops, the Internet, and mobile devices. Research topics include: large-scale video/image annotation/tagging, ranking and re-ranking, new multimedia query interfaces, multimedia search result organization and presentation, as well as accessible multimedia search.

Multimedia Advertising

The internet has become the biggest and most powerful computing and service platform for multimedia content. This content has become the primary source for more effective online advertising. Conventional advertising systems treat multimedia content the same as general text, without considering automatically monetizing the rich content of images and videos. This research direction will leverage content analysis and understanding to enable more effective and efficient advertising on multimedia content.

Mobile Multimedia Computing

Mobile devices are becoming the most frequently used terminal to access information through the Internet and social networks. More and more multimedia content is spreading over the network of mobile devices. This area is concerned with intelligent multimedia techniques to facilitate effort-free multimedia experiences on mobile devices, including media acquisition, editing, sharing, browsing, management, search, advertising, and related user interface.

Internet-scale Multimedia Retrieval and Mining

Recent years have witnessed an explosive growth of online multimedia data. Stimulated by recent progress in scalable machine learning, feature indexing and multi-modal analysis techniques, we are particularly interested in exploring challenges and new opportunities for developing internet-scale approaches for multimedia retrieval and mining. Our topics of interest are: (i) Indexing and retrieval for large multimedia collections; (ii) web-scale social- and content-network analysis; (iii) near-duplicate detection over large data sets; (iv) automatic machine tagging, semantic annotation and object recognition on massive multimedia collections; (v) scalable and distributed machine learning and data mining methods for multimedia; (vi) construction of standard large-scale multimedia collections; (vi) interfaces for exploring, browsing and visualizing large multimedia collections; and (vii) video event and temporal analysis over diverse sources.

eHealth

Our eHealth project aims to build a user-centric & data-driven personalized eHealth software and services platform for improving quality of life and efficiency of treatment. eHealth will research how personal health data can be ubiquitously collected through body sensor networks into heterogeneous devices and into the cloud; how the data can be smartly processed, visualized, analyzed, summarized, personalized, and searched; and how eHealth computing can be optimized between devices and the cloud to provide a seamless user experience and personalized services across different devices.

Media Cloud Computing

Media content distribution, searching, editing, sharing, publishing, and delivery across different devices poses a great challenge, due to multimedia’s quality of service (QoS) requirement and the heterogeneity of devices. This project envisions seamless and ubiquitous media services across different devices with good QoS. It will explore how media data and programming will be optimized among clients and the cloud, and how media content will be distributed, searched, and delivered ubiquitously across heterogeneous devices.

Human Interactive Proofs and Web Security

We are innovating technologies by exploiting the capability difference between humans and computers to: 1) Prevent bots from pretending to be humans to interact with applications designated for human users; and 2) integrate human intelligence in a computing process to solve AI (Artificial Intelligence) problems. We are also advancing technologies to fortify Web and cloud computing security.

Program Defect Detection and Performance Improvement

We are building up novel technologies to: 1) Detect and locate program defects and vulnerabilities to ensure that programs are reliable and trustworthy; and 2) diagnose performance anomalies and improve program performance.