Entity

Time filter

Source Type

Engineering, United Kingdom

Hensman J.,University of Sheffeld | Matthews A.G.,University of Cambridge | Ghahramani Z.,University of Cambridge
Journal of Machine Learning Research | Year: 2015

Gaussian process classification is a popular method with a number of appealing properties. We show how to scale the model within a variational inducing point framework, outperforming the state of the art on benchmark datasets. Importantly, the variational formulation can be exploited to allow classification in problems with millions of data points, as we demonstrate in experiments. Copyright 2015 by the authors. Source


Su D.,University of Tokyo | Nakano K.,University of Tokyo | Zheng R.,University of Tokyo | Cartmell M.P.,University of Sheffeld
International Journal of Structural Stability and Dynamics | Year: 2014

The recent potential benefit of nonlinearity has been applying in order to improve the effectiveness of energy harvesting devices. For instance, at relatively high excitation levels, both low and high-energy responses can coexist for the same parameter combinations in a hardening type Duffing oscillator, and this provides a wider bandwidth and a higher energy harvesting effectiveness under periodic excitations. However, frequency or amplitude sweeps of the excitation must be used in order to reach a desirable high-energy orbit, and this gives a limitation on practical implementation. This paper presents a stifness tunable nonlinear vibrational energy harvester which contains a moving magnetic end mass attached to a cantilever beam, whose nonlinearity emerges from the interaction forces with two neighboring permanent magnets facing with opposing poles. The motivating hypothesis has been that the jump from the low-energy orbit to the high-energy orbit can be triggered by tuning the stifness of the system without changing the frequency or the amplitude of the excitation. Theoretical investigations show a methodology for tuning stifness, and experimental tests have validated that theproposed method can be used to trigger a jump to the desirable state, and hereby this can broaden the bandwidth of the energy harvester. © 2014 World Scientific Publishing Company. Source


Mazumdar S.,University of Sheffeld | Kauppinen T.,University of Bremen | Kauppinen T.,Aalto University
CEUR Workshop Proceedings | Year: 2014

Visual exploration of data enables users and analysts observe interesting patterns that can trigger new research for further investigation. With the increasing availability of Linked Data, facilitating support for making sense of the data via visual exploration tools for hypothesis generation is critical. Time and space play important roles in this because of their ability to illustrate dynamicity, from a spatial context. Yet, Linked Data visualization approaches typically have not made efficient use of time and space together, apart from typical rather static multi-visualization approaches and mashups. In this paper we demonstrate ELBAR explorer that visualizes a vast amount of scientific observational data about the Brazilian Amazon Rainforest. Our core contribution is a novel mechanism for animating between the different observed values, thus illustrating the observed changes themselves. Source


Salman N.,University of Sheffeld | Alsindi N.,Eisala Briish Telecom Innovaion Cener EBTIC | Mihaylova L.,University of Sheffeld | Kemp A.H.,University of Leeds
2014 Workshop on Sensor Data Fusion: Trends, Solutions, Applications, SDF 2014 | Year: 2014

In this paper, we present a complete framework for accurate indoor positioning and tracking using the 802.11a WiFi network. Channel frequency response is first estimated via the least squares (LS) method using an orthogonal frequency division multiplexing (OFDM) pilot symbol. For accurate time of arrival (ToA) distance estimates in multipath environments, super resolution technique i.e. Multiple Signal Classification (MUSIC) is used which capitalizes on the autocorrelation matrix of the estimated channel frequency response. The estimated distances from the base stations (BSs) are then used in the observation model for particle filter (PF) tracking for which a constant velocity motion model is used, depicting indoor mobile movement. The tracking performance of the combined MUSIC-PF is compared with PF performance when a conventional cross correlator (CC) is used for delay estimates. It is shown via simulation that the MUSIC-PF performance is superior to the CC-PF performance. © 2014 IEEE. Source


Mazumdar S.,University of Sheffeld | Lanfranchi V.,University of Sheffeld | Cano A.E.,Open University Milton Keynes | Ciravegna F.,University of Sheffeld
CEUR Workshop Proceedings | Year: 2013

Advancements in mobile technology and the proliferation of social media platforms have made it possible for individuals to stay constantly connected with friends and family. This has provided new opportunities to the emergency response domain, where the information shared by individuals in crisis can provide invaluable insight into the situation on the ground. Information shared on social media is highly dynamic, heterogeneous, large scale, geographically distributed and multilingual. Moreover, the context of such information is mostly relevant for a very short period of time and the information can be very subjective, embedded in personal feelings. This is a significant challenge for the emergency response domain, where critical decisions need to be made quickly on the basis of the users' situation awareness. We propose to address this issue using visual analytics techniques to facilitate browsing and understanding of topicality and feelings in social media. Our approach is twofoldrstly, we enrich social media posts by adding semantics to facilitate browsing and sentiment in order to gauge the emotions behind individual posts. Secondly, we combine two paradigms of data browsing - topical and temporal into a real-time dynamic visualisation of social media messages. Source

Discover hidden collaborations