Catálogo de publicaciones - libros

In this paper we propose a novel scheme for unsupervised detection of structure in activity data. Our method is based upon an algorithm that represents data in terms of multiple low-dimensional eigenspaces. We describe the algorithm and propose an extension that allows to handle multiple time scales. The validity of the approach is demonstrated on several data sets and using two types of acceleration features. Finally, we report on experiments that indicate that our approach can yield recognition rates comparable to other, supervised approaches.

Sensor networks hold the promise of truly intelligent buildings: buildings that adapt to the behavior of their occupants to improve productivity, efficiency, safety, and security. To be practical, such a network must be economical to manufacture, install and maintain. Similarly, the methodology must be efficient and must scale well to very large spaces. Finally, be be widely acceptable, it must be inherently privacy-sensitive. We propose to address these requirements by employing networks of passive infrared (PIR) motion detectors. PIR sensors are inexpensive, reliable, and require very little bandwidth. They also protect privacy since they are neither capable of directly identifying individuals nor of capturing identifiable imagery or audio. However, with an appropriate analysis methodology, we show that they are capable of providing useful contextual information. The methodology we propose supports scalability by adopting a hierarchical framework that splits computation into localized, distributed tasks. To support our methodology we provide theoretical justification for the method that grounds it in the action recognition literature. We also present quantitative results on a dataset that we have recorded from a 400 square meter wing of our laboratory. Specifically, we report quantitative results that show better than 90% recognition performance for low-level activities such as walking, loitering, and turning. We also present experimental results for mid-level activities such as visiting and meeting.

In this paper we present a social and technical architecture which will enable the study of localization from the perspective of crowds. Our research agenda is to leverage new computing opportunities that arise when many people are simultaneously localizing themselves. By aggregating this and other types of context information we intend to develop a statistically powerful data set that can be used by urban planners, users and their software. This paper presents an end-to-end strategy, motivated with preliminary user studies, for lowering the social and technical barriers to sharing context information. The primary technology through which we motivate participation is an intelligent context-aware instant messaging client called Nomatic*Gaim. We investigate social barriers to participation with a small informal user study evaluating automatic privacy mechanisms which give people control over their context disclosure. We then analyze some preliminary data from an early deployment. Finally we show how leveraging these mass-collaborations could help to improve Nomatic*Gaim by allowing it to infer position to place mappings.

Several approaches to context awareness have been proposed ranging from unsupervised learning to ontologies. Independent of the type of context awareness used a consistent approach to naming contexts is required. A novel paradigm for labeling contexts is described based on close range wireless connections between devices and a very simple, unsupervised learning algorithm. It is shown by simulation analysis that it is possible to achieve a labeling of different contexts which allows context related information to be communicated in a consistent manner between devices. As the learning is unsupervised no user input is required for it to work. Furthermore this approach requires no extra infrastructure or resources to manage the names assigned to the contexts.

We describe an application used to share context and build common ground between nearby users. Our application runs on mobile devices and allows users securely to exchange the contents of their address books. This exchange reveals only which entries are common to the two users. We explore the use of our application using both Bluetooth and NFC as an underlying technology. Finally, we present the results of a small user study we have conducted.

Evaluating the performance of a continuous activity recognition system can be a challenging problem. To-date there is no widely accepted standard for dealing with this, and in general methods and measures are adapted from related fields such as speech and vision. Much of the problem stems from the often imprecise and ambiguous nature of the real-world events that an activity recognition system has to deal with. A recognised event might have variable duration, or be shifted in time from the corresponding real-world event. Equally it might be broken up into smaller pieces, or joined together to form larger events. Most evaluation attempts tend to smooth over these issues, using “fuzzy” boundaries, or some other parameter based error decision, so as to make possible the use of standard performance measures (such as insertions and deletions.) However, we argue that reducing the various facets of a activity system into limited error categories – that were originally intended for different problem domains – can be overly restrictive. In this paper we attempt to identify and characterise the errors typical to continuous activity recognition, and develop a method for quantifying them in an unambiguous manner. By way of an initial investigation, we apply the method to an example taken from previous work, and discuss the advantages that this provides over two of the most commonly used methods.

Context-based applications are supposed to decrease human-machine interactions. To this end, they must interpret the meaning of context data. Ontologies are a commonly accepted approach of specifying data semantics and are thus considered a precondition for the implementation of context-based systems. Yet, experiences gained from the European project Daidalos evoke concerns that this approach has its flaws when the application domain can hardly be delimited. These concerns are raised by the human limitation in dealing with complex specifications. This paper proposes a relaxation of the situation: Humans strength is the understating of natural languages, computers, however, possess superior pattern matching power. Therefore, it is suggested to enrich or even replace semantic specifications of context data items by free-text descriptions. For instance, rather than using an Ontology specification to describe an Italian restaurant the restaurant can simply be described by its menu card. To facilitate this methodology, context documents are introduced and a novel information retrieval approach is elucidated, evaluated, and analysed with the help of Bose-Einstein statistics. It is demonstrated that the new approach clearly outperforms conventional information retrieval engines and is an excellent addition to context Ontologies.

Our homes and lives are as individual as ourselves. Many aspects, such as technical equipment, furniture, and usage patterns in these surroundings differ. Thus, personalization of applications that operate in such environments is required. The challenge for tools and programming paradigms is to provide a powerful but yet easy-to-use platform. In this paper we illustrate how our visual scripting language puts these requirements for programming ubiquitous computing environments into action.