Automatic feature selection for context recognition in mobile devices

Könönen, Ville; Mäntyjärvi, Jani; Similä, Heidi; Pärkkä, Juha; Ermes, Miikka

doi:10.1016/j.pmcj.2009.07.001

Cited by 46 publications

(31 citation statements)

References 16 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Several classifications of human activities include more complex activities related with the field of sports: lying, Nordic walking, outdoor bicycling, rowing with the rowing machine, running, sitting, soccer playing, standing, walking in [45], or for daily activities such as having lunch, breakfast or dinner, going to work, shopping, sleeping, using a computer, and working [46]. The most comprehensive classification of human activities is the Physical activity compendium [35], which has 21 categories of activities as follows: Bicycling; Conditioning Exercise; Dancing, Fishing and Hunting; Home Activities; Home Repair; Inactivity; Lawn and Garden; Miscellaneous; Music Playing; Occupation; Running; Self Care; Sexual Activity; Sports; Transportation; Walking; Water Activities; Winter Activities; Religious Activities; and Volunteer Activities.…”

Section: Taxonomy Of Human Activitiesmentioning

confidence: 99%

Smartphone User Identity Verification Using Gait Characteristics

et al. 2016

View full text Add to dashboard Cite

Smartphone-based biometrics offers a wide range of possible solutions, which could be used to authenticate users and thus to provide an extra level of security and theft prevention. We propose a method for positive identification of smartphone user's identity using user's gait characteristics captured by embedded smartphone sensors (gyroscopes, accelerometers). The method is based on the application of the Random Projections method for feature dimensionality reduction to just two dimensions. Then, a probability distribution function (PDF) of derived features is calculated, which is compared against known user PDF. The Jaccard distance is used to evaluate distance between two distributions, and the decision is taken based on thresholding. The results for subject recognition are at an acceptable level: we have achieved a grand mean Equal Error Rate (ERR) for subject identification of 5.7% (using the USC-HAD dataset). Our findings represent a step towards improving the performance of gait-based user identity verification technologies.

show abstract

Section: Taxonomy Of Human Activitiesmentioning

confidence: 99%

Smartphone User Identity Verification Using Gait Characteristics

et al. 2016

View full text Add to dashboard Cite

show abstract

“…By asking the user to check and even correct the misclassified results, the classification parameters are adapted to user characteristics during an online learning (e.g. [Könönena et al, 2010] or [Brezmes et al, 2009]). Community-Guided Learning (GCL) [Peebles et al, 2010] is another available approach for generalizing classification methods.…”

Section: Learning Techniquesmentioning

confidence: 99%

“…Typically, it is preferable to use as few features as possible in mobile phone applications. This is because of two reasons: first the computational burden of feature extractions as the number of features increases , and second, the risk of obtaining suboptimal results due to classifier confusion when too many features have been used [Könönena et al, 2010]. While appropriate sensors are selected in the sensing stage, feature selection can confine the features to the most informative ones for a given sensor and the available classification technique.…”

Section: Feature Selectionmentioning

confidence: 99%

A survey on smartphone-based systems for opportunistic user context recognition

2013

View full text Add to dashboard Cite

The ever growing computation and storage capability of mobile phones has given rise to mobile centric context recognition systems, which are able to sense and analyze the context of the carrier so as to provide an appropriate level of service. Particularly, as nonintrusive autonomous sensing and context recognition are one of the most desirable characteristics of a personal sensing system; commendable efforts have been made to develop opportunistic sensing techniques on mobile phones. The resulting combination of these approaches has ushered in a new realm of applications, namely opportunistic user context recognition with mobile phones. This article surveys the existing research and approaches towards realization of such systems. In doing so, the typical architecture of a mobile centric user context recognition system as a sequential process of sensing, pre-processing and context recognition phases are introduced. The main techniques used for the realization of the respective processes during these phases are described, highlighting strengths and limitations of those. In addition, lessons learned from previous approaches are presented as motivation for future research. Finally, several open challenges are discussed as possible ways to extend the capabilities of current systems and improve their real-world experience.

show abstract

“…Consideration must be given to the utility of each feature introduced to the system. A common approach to this problem is to generate a wide range of features, and to then allow the system to select those which are most helpful in classifying the data [13]. This approach is used in this architecture, as covered in the next section.…”

Section: Feature Extractionmentioning

confidence: 99%

“…Many techniques use supervised methods, in which training data is collected and manually marked with the correct context. From this information, models can be built using a variety of statistical methods [7,13] . Alternatively, unsupervised learning recognises emerging patterns in data without manually marked training data.…”

Section: Introductionmentioning

confidence: 99%

Adaptive sampling in context-aware systems: a machine learning approach

Wood¹,

Merrett²,

Gunn³

et al. 2012

IET Conference on Wireless Sensor Systems (WSS 2012)

View full text Add to dashboard Cite

As computing systems become ever more pervasive, there is an increasing need for them to understand and adapt to the state of the environment around them: that is, their context. This understanding comes with considerable reliance on a range of sensors. However, portable devices are also very constrained in terms of power, and hence the amount of sensing must be minimised. In this paper, we present a machine learning architecture for context awareness which is designed to balance the sampling rates (and hence energy consumption) of individual sensors with the significance of the input from that sensor. This significance is based on predictions of the likely next context. The architecture is implemented using a selected range of user contexts from a collected data set. Simulation results show reliable context identification results. The proposed architecture is shown to significantly reduce the energy requirements of the sensors with minimal loss of accuracy in context identification.

show abstract

Automatic feature selection for context recognition in mobile devices

Cited by 46 publications

References 16 publications

Smartphone User Identity Verification Using Gait Characteristics

Smartphone User Identity Verification Using Gait Characteristics

A survey on smartphone-based systems for opportunistic user context recognition

Adaptive sampling in context-aware systems: a machine learning approach

Contact Info

Product

Resources

About