291 research outputs found
User Activity Related Data Sets for Context Recognition
No full textAbstract. The use of body-worn sensors for recognizing a person’s context has gained much popularity recently. For the development of suitable context recognition approaches and their evaluation, real-world data is essential. In this paper, we present two data sets which we recorded to evaluate the usefulness of sensors and to develop, test and improve our recognition strategies with respect to two specific recognition tasks
The superorganism of massive collective wearables
No full textPersonalized wearable ICT systems presented in fashionable and appealing lifestyle-designs have gained critical user acceptance, and comprise momentum to bring wearable computing to a socio-technical mass phenomenon within the next few years. Early indicators for this expected wearable systems "tsunami" are the "spring tide" of 5.3 billion mobile phone platforms (i.e. mobile subscribers) as of the end of 2013, an assessed market potential for 300 million smart watches in 2014, and a possible market for more than 200 million smart eye-wear systems in 2015 [1].
This workshop asks the questions on the potentials and opportunities of turning these massively deployed wearable systems to a globe spanning super-organism of socially interactive personal digital assistants. While the individual wearables are of heterogeneous provenance and typically act autonomously, we can assume that they can (and will) self-organize into large scale cooperative collectives, with humans being mostly out-of-the-loop [2]. We may not assume a common objective or central controller, but rather volatile network topologies, co-dependence and internal competition, non-linear and non-continuous dynamics, and sub-ideal, failure prone operation. We could refer to these emerging massive collectives of wearables as a "super-organism" [7], since it exhibits properties of a living organism (like e.g. 'collective intelligence') on its own. In order to properly exploit such super-organisms, we need to develop a deeper scientific understanding of the foundational principles by which they operate
ACOMORE'14: The first symposium on activity and context modeling and recognition, 2014 - Welcome and committees
No full textExplainable for Trustworthy AI
No full textBlack-box Artificial Intelligence (AI) systems for automated decision making are often based on over (big) human data, map a user’s features into a class or a score without exposing why. This is problematic for the lack of transparency and possible biases inherited by the algorithms from human prejudices and collection artefacts hidden in the training data, leading to unfair or wrong decisions. The future of AI lies in enabling people to collaborate with machines to solve complex problems. This requires good communication, trust, clarity, and understanding, like any efficient collaboration. Explainable AI (XAI) addresses such challenges, and for years different AI communities have studied such topics, leading to different definitions, evaluation protocols, motivations, and results. This chapter provides a reasoned introduction to the work of Explainable AI to date and surveys the literature focusing on symbolic AI-related approaches. We motivate the needs of XAI in real-world and large-scale applications while presenting state-of-the-art techniques and best practices and discussing the many open challenges
Socio-inspired ICT: Towards a socially grounded society-ICT symbiosis
No full textModern ICT (Information and Communication Technology) has developed a vision where the “computer” is no longer associated with the concept of a single device or a network of devices, but rather the entirety of situated services originating in a digital world, which are perceived through the physical world. It is observed that services with explicit user input and output are becoming to be replaced by a computing landscape sensing the physical world via a huge variety of sensors, and controlling it via a plethora of actuators. The nature and appearance of computing devices is changing to be hidden in the fabric of everyday life, invisibly networked, and omnipresent, with applications greatly being based on the notions of context and knowledge. Interaction with such globe spanning, modern ICT systems will presumably be more implicit, at the periphery of human attention, rather than explicit, i.e. at the focus of human attention.Socio-inspired ICT assumes that future, globe scale ICT systems should be viewed as social systems. Such a view challenges research to identify and formalize the principles of interaction and adaptation in social systems, so as to be able to ground future ICT systems on those principles. This position paper therefore is concerned with the intersection of social behaviour and modern ICT, creating or recreating social conventions and social contexts through the use of pervasive, globe-spanning, omnipresent and participative ICTValues and TechnologyTechnology, Policy and Managemen
Evaluating Performance in Continuous Context Recognition Using Event-Driven Error Characterisation
No full textEvaluating 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 âÂ�Â�fuzzyâÂ�Â�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
- …
