Nitya Narasimhan

Interactive visual object search through mutual information maximization

Searching for small objects (e.g., logos) in images is a critical yet challenging problem. It becomes more difficult when target objects differ significantly from the query object due to changes in scale, viewpoint or style, not to mention partial occlusion or cluttered backgrounds. With the goal to retrieve and accurately locate the small object in the images, we formulate the object search as the problem of finding subimages with the largest mutual information toward the query object. Each image is characterized by a collection of local features. Instead of only using the query object for matching, we propose a discriminative matching using both positive and negative queries to obtain the mutual information score. The user can verify the retrieved subimages and improve the search results incrementally. Our experiments on a challenging logo database of 10,000 images highlight the effectiveness of this approach.

A Lightweight Remote Display Management Protocol for Mobile Devices

Mobile phone users rely on their device displays to provide both information (e.g., currently-playing song) and alerts (e.g., caller ID). Unfortunately, phone inaccessibility coupled with display or user constraints often make it difficult for users to view and respond to alerts in a timely manner. Our lightweight remote display (LRD) concept addresses this problem by enabling phones to discover and use alternative displays nearby. The LRD management protocol (LRDMP) is specifically tuned for use with resource-constrained displays such as wrist-watches, yet designed to be extensible for use with a wide variety of display devices with diverse capabilities. In this paper, we describe the design and implementation of LRDMP, discuss our prototype deployment model and conclude with a look at related work and relevant applications of this technology.

Investigating the potential of in-home devices for improving medication adherence

We conducted five focus groups with seniors and middle-aged participants who live independently in their own homes to assess the potential value of a home-centered medication reminder system concept. The medication reminder system was conceptualized as a system that uses a television and set-top box, mobile phones and other in-home accessories as a means to set and deliver medication reminders. We found that the main value perceived by participants in the medication reminder system was its ability to provide multiple channels for them to be reminded of medications. The mobile phone, due to its advantages in portability and privacy, was considered to be the most useful device on which to receive reminders. Most participants saw value in receiving secondary reminders on other devices in their home such as the TV, PC, and other in-home accessories. Design implications along with other findings about the challenges faced by participants in managing their medications are discussed.

Is this urgent?: exploring time-sensitive information needs in collaborative question answering

As online Collaborative Question Answering (CQA) servicessuch as Yahoo! Answers and Baidu Knows are attracting users, questions, and answers at an explosive rate, the truly urgent and important questions are increasingly getting lost in the crowd. That is, questions that require immediate responses are pushed out of the way by the trivial but more recently arriving questions. Unlike other questions in collaborative question answering (CQA) for which users might be willing to wait until good answers appear, urgent questions are likely to be of interest to the asker only if answered in the next few minutes or hours. For such questions, late responses are either not useful or are simply not applicable. Unfortunately, current collaborative question-answering systems do not distinguish urgent questions from the rest, and could thus be ineffective for urgent information needs. We explore text- and data- mining methods for automatically identifying urgent questions in the CQA setting. Our results indicate that modeling the question context (i.e., the particular forum/category where the question was posted) can increase classification accuracy compared to the text of the question alone.

When the Shift Hits the (Television) Fan: A Growing Opportunity for Companion Devices

The television ecosystem faces growing attention and audience fragmentation thanks to an explosion of content sources and content consumption devices. In this article, the author looks at some of the interaction paradigms, viewer behaviors, and innovative social and dual-screen experiences that seek to disrupt or redefine the traditional models of audience measurement and content distribution, in an effort to address these challenges.

VStore: efficiently storing virtualized state across mobile devices

Mobile virtualization is a nascent technology the value of which lies in enforcing data protection, providing process isolation and simplifying application reuse across device platforms. In this paper, we present VStore, a flexible mechanism for storage management and content protection that exploits virtualization to modularize data access and sharing mechanisms into containers separate from those containing guest operating systems and applications. This can provide mobile devices with rich storage options, including local, remote, or peer to peer stores, without affecting guest operating systems, middleware, or applications, and it enables diverse content create-query-share semantics. Further, VStore can provide the means to support centralized content protection and access control to all resident applications, thereby enabling new content distribution or privacy preservation policies to be enforced transparently. We discuss initial VStore implementation results and conclude by outlining new opportunities and challenges for further research.

Efficient indexing structure for scalable processing of spatial alarms

We present the design and implementation of a new indexing technique, Mondrian tree. The Mondrian tree indexing method partitions the entire universe of discourse into spatial alarm monitoring regions and alarm-free regions. This enables us to reduce the number of on-demand alarm-free region computations, significant savings of both server load and client-to-server communication cost. We evaluate the efficiency of the Mondrian tree indexing approach and show that the Mondrian tree offers significant performance enhancements on spatial alarm processing at both the server side and the client side.

Middleware for pervasive and ad hoc computing

More than a decade has passed since Mark Weiser articulated his vision of a world of pervasive devices, leveraging contextual information, to provide an increasingly interactive and responsive environment to users, to aid them in their everyday activities [1]. However, despite considerable progress, the promise of pervasive computing still remains elusive. The diversity in currently available devices, networking infrastructure and information content has complicated Weisers vision, forcing many current research and development projects to focus only on point-examples of this technology. The organisation of the workshop on Middleware for Pervasive and Ad-Hoc Computing (MPAC) has been motivated by our belief that underpinning middleware mechanisms are central in weaving together the multitude of computing, communication and information technologies. Middleware for pervasive computing and ad hoc networking provide two core research areas in this respect. In particular, pervasive computing middleware will allow you to take advantage of the resources in your environment to tailor your services and applications for seamless access and unrestricted mobility. Ad hoc networking middleware will permit the formation of ad hoc communities for new applications. However, existing middleware technologies do not fully exploit the potential of such pervasive and ad hoc environments.

MobiMan: Bringing Scripted Agents to Wireless Terminal Management

The increasing software complexity of wireless devices and wireless data service provisioning motivates a wireless terminal management challenge. The systems management solution for this problem needs to scale up to large device populations, while being lightweight enough to be pragmatic for resource-constrained devices. The work in this paper builds upon the emerging SyncML standard for wireless terminal management in order to bring sophisticated policy-based management to large populations of wireless data devices. It is anticipated that this technology will simplify the upgrade and management of wireless data devices substantially, thus encouraging the adoption of sophisticated data terminals.