Proceedings of the 2015 ACM International Symposium on Wearable Computers - ISWC '15, 2015
Since people perform activities differently, to avoid overfitting, creating a general model with ... more Since people perform activities differently, to avoid overfitting, creating a general model with activity data of various users is required before the deployment for personal use. However, annotating a large amount of activity data is expensive and time-consuming. In this paper, we create a general model for activity recognition with a limited amount of labelled data. We combine Latent Dirichlet Allocation (LDA) and AdaBoost to jointly train a general activity model with partially labelled data. After that, when AdaBoost is used for online prediction, we combine it with graphical models (such as HMM and CRF) to exploit the temporal information in human activities to smooth out accidental misclassifications. Experiments on publicly available datasets show that we are able to obtain the accuracy of more than 90% with 1% labelled data.
ABSTRACT Increasing interest in ontologies in the last few years has led to a shift in the resear... more ABSTRACT Increasing interest in ontologies in the last few years has led to a shift in the research focus of the context-awareness community. This shift reflects the potential of ontology-based approaches to improve upon previ-ously used context modelling approaches by ...
Fourth International Conference on Information Communications and Signal Processing 2003 and the Fourth Pacific Rim Conference on Multimedia Proceedings of the 2003 Joint, 2003
ABSTRACT This paper explores the concept of parallel TCP, which splits a traditional single TCP c... more ABSTRACT This paper explores the concept of parallel TCP, which splits a traditional single TCP connection into a number of virtual ones in parallel. A performance model of parallel TCP is presented to explain how parallel TCP can improve throughput performance. The paper also discusses the factors that affect the performance improvement, including number of connections, packet loss rate, and congestion avoidance. Simulations validate the model.
On the Move to Meaningful Internet Systems 2005 Otm 2005 Workshops Proceedings, 2005
The immaturity of the field of context-aware computing means that little is known about how to in... more The immaturity of the field of context-aware computing means that little is known about how to incorporate appropriate personalisation mechanisms into context-aware applications. One of the main challenges is how to elicit and represent complex, context-dependent requirements, and then use the resulting representations within context-aware applica- tions to support decision-making processes. In this paper, we characterise several approaches to personalisation
To develop robust, evolvable, context-aware applications that are e xible enough to recongure and... more To develop robust, evolvable, context-aware applications that are e xible enough to recongure and adapt in response to changes in context such as location, connectivity, resources, security, and user preferences, a rich context model, infras- tructure components, and programming paradigms are re- quired. The development and deployment of context-aware applications can be further assisted by tools to automate the transformation of the context model into appropriate forms for ready manipulation by developers and administrators. This paper describes an extensible set of tools we are devel- oping for use in conjunction with our existing context mod- elling framework and infrastructure that provides substantial assistance to the development and deployment of context- aware applications.
ABSTRACT Wireless networks such as Bluetooth, WLAN and WiMax have transformed the way we access i... more ABSTRACT Wireless networks such as Bluetooth, WLAN and WiMax have transformed the way we access information and communicate seamlessly whether we are at home, in the office, or on the move on a train, bus or even aircraft. As mobile and embedded computing devices become more omnipresent, it will become increasingly difficult to interconnect them via wires and single-hop wireless links limited by radio transmission range. This has given rise to mobile ad hoc networks (MANET) where far away nodes communicate by requesting intermediate nodes to relay their information in order to reach the destination. MANETs self-organize, self-configure and self-heal themselves. MANETs are being used in many applications ranging from emergency response situations to wireless vehicular ad hoc networks. Many applications of MANETs such as Emergency Response and First Responders have strict Quality of Service (QoS) requirements for their communications systems, making MANET QoS provisioning mechanisms very crucial for supporting multimedia communications such as real-time audio and video. However, QoS provisioning in MANETs is quite tough in comparison to QoS provisioning in wireline IP networks. This is due to numerous reasons such as the dynamic network topology, unpredictable communication medium and limited battery power of mobile devices forming the network. This chapter describes the challenges and the current state of the art of QoS protocols and mechanisms in MANETs.
While others have attempted to determine, by way of mathematical formulae, optimal resource dupli... more While others have attempted to determine, by way of mathematical formulae, optimal resource duplication strategies for random walk protocols, this paper is concerned with studying the emergent effects of dynamic resource propagation and replication. In particular, we show, via modelling and experimentation, that under any given decay (purge) rate the number of nodes that have knowledge of particular resource converges to a fixed point or a limit cycle. We also show that even for high rates of decay–that is, when few nodes have ...
... X Organization Lars Erik Holmquist, Viktoria Institute, Sweden Elaine Huang, RWTH Aachen Univ... more ... X Organization Lars Erik Holmquist, Viktoria Institute, Sweden Elaine Huang, RWTH Aachen University, Germany Stephen Intille, MIT, USA ... 38 Joe Marshall, Tony Pridmore, Mike Pound, Steve Benford, and Boriana Koleva Applications for Mobile Devices MakeIt: Integrate User ...
Proceedings of the 2015 ACM International Symposium on Wearable Computers - ISWC '15, 2015
Since people perform activities differently, to avoid overfitting, creating a general model with ... more Since people perform activities differently, to avoid overfitting, creating a general model with activity data of various users is required before the deployment for personal use. However, annotating a large amount of activity data is expensive and time-consuming. In this paper, we create a general model for activity recognition with a limited amount of labelled data. We combine Latent Dirichlet Allocation (LDA) and AdaBoost to jointly train a general activity model with partially labelled data. After that, when AdaBoost is used for online prediction, we combine it with graphical models (such as HMM and CRF) to exploit the temporal information in human activities to smooth out accidental misclassifications. Experiments on publicly available datasets show that we are able to obtain the accuracy of more than 90% with 1% labelled data.
ABSTRACT Increasing interest in ontologies in the last few years has led to a shift in the resear... more ABSTRACT Increasing interest in ontologies in the last few years has led to a shift in the research focus of the context-awareness community. This shift reflects the potential of ontology-based approaches to improve upon previ-ously used context modelling approaches by ...
Fourth International Conference on Information Communications and Signal Processing 2003 and the Fourth Pacific Rim Conference on Multimedia Proceedings of the 2003 Joint, 2003
ABSTRACT This paper explores the concept of parallel TCP, which splits a traditional single TCP c... more ABSTRACT This paper explores the concept of parallel TCP, which splits a traditional single TCP connection into a number of virtual ones in parallel. A performance model of parallel TCP is presented to explain how parallel TCP can improve throughput performance. The paper also discusses the factors that affect the performance improvement, including number of connections, packet loss rate, and congestion avoidance. Simulations validate the model.
On the Move to Meaningful Internet Systems 2005 Otm 2005 Workshops Proceedings, 2005
The immaturity of the field of context-aware computing means that little is known about how to in... more The immaturity of the field of context-aware computing means that little is known about how to incorporate appropriate personalisation mechanisms into context-aware applications. One of the main challenges is how to elicit and represent complex, context-dependent requirements, and then use the resulting representations within context-aware applica- tions to support decision-making processes. In this paper, we characterise several approaches to personalisation
To develop robust, evolvable, context-aware applications that are e xible enough to recongure and... more To develop robust, evolvable, context-aware applications that are e xible enough to recongure and adapt in response to changes in context such as location, connectivity, resources, security, and user preferences, a rich context model, infras- tructure components, and programming paradigms are re- quired. The development and deployment of context-aware applications can be further assisted by tools to automate the transformation of the context model into appropriate forms for ready manipulation by developers and administrators. This paper describes an extensible set of tools we are devel- oping for use in conjunction with our existing context mod- elling framework and infrastructure that provides substantial assistance to the development and deployment of context- aware applications.
ABSTRACT Wireless networks such as Bluetooth, WLAN and WiMax have transformed the way we access i... more ABSTRACT Wireless networks such as Bluetooth, WLAN and WiMax have transformed the way we access information and communicate seamlessly whether we are at home, in the office, or on the move on a train, bus or even aircraft. As mobile and embedded computing devices become more omnipresent, it will become increasingly difficult to interconnect them via wires and single-hop wireless links limited by radio transmission range. This has given rise to mobile ad hoc networks (MANET) where far away nodes communicate by requesting intermediate nodes to relay their information in order to reach the destination. MANETs self-organize, self-configure and self-heal themselves. MANETs are being used in many applications ranging from emergency response situations to wireless vehicular ad hoc networks. Many applications of MANETs such as Emergency Response and First Responders have strict Quality of Service (QoS) requirements for their communications systems, making MANET QoS provisioning mechanisms very crucial for supporting multimedia communications such as real-time audio and video. However, QoS provisioning in MANETs is quite tough in comparison to QoS provisioning in wireline IP networks. This is due to numerous reasons such as the dynamic network topology, unpredictable communication medium and limited battery power of mobile devices forming the network. This chapter describes the challenges and the current state of the art of QoS protocols and mechanisms in MANETs.
While others have attempted to determine, by way of mathematical formulae, optimal resource dupli... more While others have attempted to determine, by way of mathematical formulae, optimal resource duplication strategies for random walk protocols, this paper is concerned with studying the emergent effects of dynamic resource propagation and replication. In particular, we show, via modelling and experimentation, that under any given decay (purge) rate the number of nodes that have knowledge of particular resource converges to a fixed point or a limit cycle. We also show that even for high rates of decay–that is, when few nodes have ...
... X Organization Lars Erik Holmquist, Viktoria Institute, Sweden Elaine Huang, RWTH Aachen Univ... more ... X Organization Lars Erik Holmquist, Viktoria Institute, Sweden Elaine Huang, RWTH Aachen University, Germany Stephen Intille, MIT, USA ... 38 Joe Marshall, Tony Pridmore, Mike Pound, Steve Benford, and Boriana Koleva Applications for Mobile Devices MakeIt: Integrate User ...
Uploads
Papers by Jadwiga Indulska