As computing moves beyond the desktop, human activity becomes an essential component of many applications. Activity classification is an active research area and several research systems have been constructed. Most have focused on fragile custom hardware only available in limited quantities. We instead seek to use commodity hardware to lower the barrier to creating activity-informed mobile applications. We describe iLearn, our system for classifying human activities using the Apple iPhone‟s three-axis accelerometer and the Nike+iPod Sport Kit. Our results suggest activities including running, walking, bicycling, and sitting can be recognized at accuracies of 97 % without any training by an end-user.

Mobile devices are becoming increasingly sophisticated and the latest generation of smart cell phones now incorporates many diverse and powerful sensors. These sensors include GPS sensors, vision sensors (i.e., cameras), audio sensors (i.e., microphones), light sensors, temperature sensors, direction sensors (i.e., magnetic compasses), and acceleration sensors (i.e., accelerometers). The availability of these sensors in mass-marketed communication devices creates exciting new opportunities for data mining and data mining applications. In this paper we describe and evaluate a system that uses phone-based accelerometers to perform activity recognition, a task which involves identifying the physical activity a user is performing. To implement our system we collected labeled accelerometer data from twenty-nine users as they performed daily activities such as walking, jogging, climbing stairs, sitting, and standing, and then aggregated this time series data into examples that summarize the user activity over 10second intervals. We then used the resulting training data to induce a predictive model for activity recognition. This work is significant because the activity recognition model permits us to gain useful knowledge about the habits of millions of users passively—just by having them carry cell phones in their pockets. Our work has a wide range of applications, including automatic customization of the mobile device’s behavior based upon a user’s activity (e.g., sending calls directly to voicemail if a user is jogging) and generating a daily/weekly activity profile to determine if a user (perhaps an obese child) is performing a healthy amount of exercise.

In this work we discover the daily location-driven routines which are contained in a massive reallife human dataset collected by mobile phones. Our goal is the discovery and analysis of human routines which characterize both individual and group behaviors in terms of location patterns. We develop an unsupervised methodology based on two differing probabilistic topic models and apply them to the daily life of 97 mobile phone users over a 16 month period to achieve these goals. Topic models are probabilistic generative models for documents that identify the latent structure that underlies a set of words. Routines dominating the entire group’s activities, identified with a methodology based on the Latent Dirichlet Allocation topic model, include “going to work late”, “going home early”, “working non-stop ” and “having no reception (phone off) ” at different times over varying time-intervals. We also detect routines which are characteristic of users, with a methodology based on the Author-Topic model. With the routines discovered, and the two methods of characterizing days and users, we can then perform various tasks. We use the routines discovered to determine behavioral patterns of users and groups of users. For example, we can find individuals that display specific daily routines, such as “going to work early ” or “turning off the mobile (or having no reception) in the evenings”. We are also able to characterize daily patterns

Abstract. The automatic analysis of real-life, long-term behavior and dynamics of individuals and groups from mobile sensor data constitutes an emerging and challenging domain. We present a framework to classify people’s daily routines (defined by day type, and by group affiliation type) from real-life data collected with mobile phones, which include physical location information (derived from cell tower connectivity), and social context (given by person proximity information derived from Bluetooth). We propose and compare single- and multi-modal routine representations at multiple time scales, each capable of highlighting different features from the data, to determine which best characterized the underlying structure of the daily routines. Using a massive data set of 87000+ hours spanning four months of the life of 30 university students, we show that the integration of location and social context and the use of multiple time-scales used in our method is effective, producing accuracies of over 80 % for the two daily routine classification tasks investigated, with significant performance differences with respect to the single-modal cues. 1

We create a performance animation system that leverages the power of low-cost accelerometers, readily available motion capture databases, and construction techniques from e-textiles. Our system, built with only off-theshelf parts, consists of five accelerometers sewn into a comfortable shirt that streams data to a computer. The accelerometer readings are continuously matched against accelerations computed from existing motion capture data, and an avatar is animated with the closest match. We evaluate our system visually and using simultaneous motion and accelerometer capture. Categories and Subject Descriptors (according to ACM CCS): I.3.7 [Computer Graphics]: Three-Dimensional Graphics and Realism – animation. I.3.6 [Computer Graphics]: Methodology and Techniques – interaction techniques.

Abstract—When introducing a novel sensor to the context recognition community, one of the major challenges is to support reproducability under similar conditions. In order to get a grasp on this process, we divide the context recognition into three subsections: the physical environment and how it is affected by the context, the sensor and how it can represent the attributes of the physical world, and the classifying method and how it deciphers the sensory representation. We then outlined our recommendation for a methodology to formally describe the context and sensor subdivisions in order to isolate and quantify error within the system. The result would develop a basis of standard models of activities and contexts within the community which would serve to improve evaluation of novel sensors and classification algorithms. I.

Monitoring human motion has recently received great attention and can be used in many applications, such as human motion prediction. We present the collected data set from a body sensor network attached to the human body. The set of sensors consists of accelerometers measuring acceleration in three directions that are attached to the upper and lower back as well as the knees and ankles. In addition, pressures on the insoles are measured with four pressure sensors inside each shoe. Two types of motion are considered: walking backwards on a straight line and walking forwards on a figure-8 path. Finally, we study and present basic statistics of the data.

Due to the rapid aging of the European population, an effort needs to be made to ensure that the elderly can live longer independently with minimal support of the working-age population. The Confidence project aims to do this by unobtrusively monitoring their activity to recognize falls and other health problems. This is achieved by equipping the user with radio tags, from which the locations of body parts are determined, thus enabling posture and movement reconstruction. In the paper we first give a general overview of the research on fall detection and activity recognition. We proceed to describe the machine learning approach to activity recognition to be used in the Confidence project. In this approach, the attributes characterizing the user’s behavior and a machine learning algorithm must be selected. The attributes we consider are the locations of body parts in the reference coordinate system (fixed with respect to the environment), the locations of body parts in a body coordinate system (affixed to the user’s body) and the angles between adjacent body parts. Eight machine learning algorithms are compared. The highest classification accuracy of over 95 % is achieved by Support Vector Machine used on the reference attributes and angles. Povzetek: Članek opisuje zaznavanje padcev in prepoznavanja aktivnosti nasploh ter izvedbo prepoznavanja aktivnosti s strojnim učenjem za potrebe projekta Confidence.

Obesity is now considered a global epidemic and is predicted to become the number one preventive health threat in the industrialized world. Presently, over 60 % of the U.S. adult population is overweight and 30 % is obese. This is of concern because obesity is linked to leading causes of death, such as heart and pulmonary diseases, stroke, and type 2 diabetes. The dramatic rise in obesity rates is attributed to an environment that provides easy access to high caloric food and drink and promotes low levels of physical activity. Unfortunately, many people have a poor understanding of their own daily energy (im)balance: the number of calories they consume from food compared with what they expend through physical activity. Accelerometers offer promise as an objective measure of physical activity. In prior work they have been used to estimate energy expenditure and activity type. This work further demonstrates how wireless accelerometers can be used for real-time automatic recognition of physical activity type, intensity, and duration and estimation of energy expenditure. The parameters of the algorithms such as type of classifier/regressor, feature set, window length, signal preprocessing, sensor set utilized

We describe SocialCircuits, a platform capable of measuring the face-to-face and phone-based communication network of a realworld community. This platform uses commodity mobile phones to measure social ties between individuals, and uses long and short term surveys to measure the shifts in individual habits, opinions, health, and friendships influenced by those ties. We also describe the flagship experiment using this platform, a year-long study of an entire university undergraduate dormitory. Lastly, we discuss some of the key challenges we met in building and deploying the platform, including mobile phone hardware and software selection, privacy considerations, community selection and recruitment, and techniques for minimizing data loss.