Clustering-Based Ensemble Learning for Activity Recognition in Smart Homes

Application of sensor-based technology within activity monitoring systems is becoming a popular technique within the smart environment paradigm. Nevertheless, the use of such an approach generates complex constructs of data, which subsequently requires the use of intricate activity recognition techniques to automatically infer the underlying activity. This paper explores a cluster-based ensemble method as a new solution for the purposes of activity recognition within smart environments. With this approach activities are modelled as collections of clusters built on different subsets of features. A classification process is performed by assigning a new instance to its closest cluster from each collection. Two different sensor data representations have been investigated, namely numeric and binary. Following the evaluation of the proposed methodology it has been demonstrated that the cluster-based ensemble method can be successfully applied as a viable option for activity recognition. Results following exposure to data collected from a range of activities indicated that the ensemble method had the ability to perform with accuracies of 94.2% and 97.5% for numeric and binary data, respectively. These results outperformed a range of single classifiers considered as benchmarks.

Kinect and Episodic Reasoning for Human Action Recognition

This paper presents a method for rational behaviour recognition that combines vision-based pose estimation with knowledge modeling and reasoning. The proposed method consists of two stages. First, RGB-D images are used in the estimation of the body postures. Then, estimated actions are evaluated to verify that they make sense. This method requires rational behaviour to be exhibited. To comply with this requirement, this work proposes a rational RGB-D dataset with two types of sequences, some for training and some for testing. Preliminary results show the addition of knowledge modeling and reasoning leads to a significant increase of recognition accuracy when compared to a system based only on computer vision.

A case study in identifying acceptable bitrates for human face recognition tasks

Face recognition from images or video footage requires a certain level of recorded image quality. This paper derives acceptable bitrates (relating to levels of compression and consequently quality) of footage with human faces, using an industry implementation of the standard H.264/MPEG-4 AVC and the Closed-Circuit Television (CCTV) recording systems on London buses. The London buses application is utilized as a case study for setting up a methodology and implementing suitable data analysis for face recognition from recorded footage, which has been degraded by compression. The majority of CCTV recorders on buses use a proprietary format based on the H.264/MPEG-4 AVC video coding standard, exploiting both spatial and temporal redundancy. Low bitrates are favored in the CCTV industry for saving storage and transmission bandwidth, but they compromise the image usefulness of the recorded imagery. In this context, usefulness is determined by the presence of enough facial information remaining in the compressed image to allow a specialist to recognize a person. The investigation includes four steps: (1) Development of a video dataset representative of typical CCTV bus scenarios. (2) Selection and grouping of video scenes based on local (facial) and global (entire scene) content properties. (3) Psychophysical investigations to identify the key scenes, which are most affected by compression, using an industry implementation of H.264/MPEG-4 AVC. (4) Testing of CCTV recording systems on buses with the key scenes and further psychophysical investigations. The results showed a dependency upon scene content properties. Very dark scenes and scenes with high levels of spatial–temporal busyness were the most challenging to compress, requiring higher bitrates to maintain useful information.

ReadingAct RGB-D action dataset and human action recognition from local features

For general home monitoring, a system should automatically interpret people’s actions. The system should be non-intrusive, and able to deal with a cluttered background, and loose clothes. An approach based on spatio-temporal local features and a Bag-of-Words (BoW) model is proposed for single-person action recognition from combined intensity and depth images. To restore the temporal structure lost in the traditional BoW method, a dynamic time alignment technique with temporal binning is applied in this work, which has not been previously implemented in the literature for human action recognition on depth imagery. A novel human action dataset with depth data has been created using two Microsoft Kinect sensors. The ReadingAct dataset contains 20 subjects and 19 actions for a total of 2340 videos. To investigate the effect of using depth images and the proposed method, testing was conducted on three depth datasets, and the proposed method was compared to traditional Bag-of-Words methods. Results showed that the proposed method improves recognition accuracy when adding depth to the conventional intensity data, and has advantages when dealing with long actions.

Effects of environmental conditions, human activity, reproduction, antler cycle and grouping on fecal glucocorticoids of free-ranging Pampas deer stags (Ozotoceros bezoarticus bezoarticus)

In this study, a commercial enzyme immunoassay (EIA) was validated in detecting glucocorticoids in Pampas deer feces, in order to investigate the influence of several factors on the adrenocortical function. Fecal samples, behavioral data and information concerning male grouping and antlers status were collected at a monthly basis during a 1 year period from free-ranging stags living at Emas National Park, Brazil (18 degrees S/52 degrees W). The results revealed that concentrations of fecal glucocorticoids in winter were significantly higher than those corresponding to spring and summer. In addition, dry season data presented higher levels than during the wet season. Significant difference was found between fecal levels of breeding stags in summer and nonbreeding stags, whereas no difference was observed between breeding stags in winter and nonbreeding stags. on the other hand, males from areas with frequent human disturbance exhibited higher glucocorticoid concentrations and flight distances than individuals from areas of lower human activity. Males with antlers in velvet had elevated levels compared with animals in hard antler or antler casting. Also, we found that glucocorticoid levels were higher in groups with three or more males than in groups with only one male. The flight distances showed positive correlation with fecal glucocorticoid. These data indicate that fecal glucocorticoid provides a useful approach in the evaluation of physiological effects of environment, inter-individuals relationship and human-induced stressors on free-ranging Pampas deer stags. (c) 2005 Elsevier B.V. All rights reserved.

The Interaction Between Forest Fires and Human Activity in Southern Switzerland

The impact of human activities on the fire regime in southern Switzerland was studied using (pre)historical charcoal and pollen data from lake sediments and statistical data from the 20th century. The cultural impact on forest fire was established by correlating charcoal-influx data with pollen percentages of anthropogenic indicators such as Plantago lanceolata, the Cerealia (sum of Avena t., Triticum t. and Hordeum t.) and Secale. During the 20th century, fire frequency was correlated with precipitation, dry and very dry periods and landscape management indicators. The effects of human activity on the fire regime are clearly recognisable since at least the Neolithic period. Using palaeoecological or statistical data, the variations in fire regime originating from anthropogenic actions may be differentiated from those due to climatic changes if they are sufficiently conspicuous.

Local Human Activity in two communities of Formosa, Argentina

author must provide abstract

Enhancing Activity Recognition by Fusing Inertial and Biometric Information

Activity recognition is an active research field nowadays, as it enables the development of highly adaptive applications, e.g. in the field of personal health. In this paper, a light high-level fusion algorithm to detect the activity that an individual is performing is presented. The algorithm relies on data gathered from accelerometers placed on different parts of the body, and on biometric sensors. Inertial sensors allow detecting activity by analyzing signal features such as amplitude or peaks. In addition, there is a relationship between the activity intensity and biometric response, which can be considered together with acceleration data to improve the accuracy of activity detection. The proposed algorithm is designed to work with minimum computational cost, being ready to run in a mobile device as part of a context-aware application. In order to enable different user scenarios, the algorithm offers best-effort activity estimation: its quality of estimation depends on the position and number of the available inertial sensors, and also on the presence of biometric information.

Towards a fuzzy-based multi-classifier selection module for activity recognition applications

Performing activity recognition using the information provided by the different sensors embedded in a smartphone face limitations due to the capabilities of those devices when the computations are carried out in the terminal. In this work a fuzzy inference module is implemented in order to decide which classifier is the most appropriate to be used at a specific moment regarding the application requirements and the device context characterized by its battery level, available memory and CPU load. The set of classifiers that is considered is composed of Decision Tables and Trees that have been trained using different number of sensors and features. In addition, some classifiers perform activity recognition regardless of the on-body device position and others rely on the previous recognition of that position to use a classifier that is trained with measurements gathered with the mobile placed on that specific position. The modules implemented show that an evaluation of the classifiers allows sorting them so the fuzzy inference module can choose periodically the one that best suits the device context and application requirements.

Mars: a personalised mobile activity recognition system

Mobile activity recognition focuses on inferring the current activities of a mobile user by leveraging the sensory data that is available on today’s smart phones. The state of the art in mobile activity recognition uses traditional classification learning techniques. Thus, the learning process typically involves: i) collection of labelled sensory data that is transferred and collated in a centralised repository; ii) model building where the classification model is trained and tested using the collected data; iii) a model deployment stage where the learnt model is deployed on-board a mobile device for identifying activities based on new sensory data. In this paper, we demonstrate the Mobile Activity Recognition System (MARS) where for the first time the model is built and continuously updated on-board the mobile device itself using data stream mining. The advantages of the on-board approach are that it allows model personalisation and increased privacy as the data is not sent to any external site. Furthermore, when the user or its activity profile changes MARS enables promptly adaptation. MARS has been implemented on the Android platform to demonstrate that it can achieve accurate mobile activity recognition. Moreover, we can show in practise that MARS quickly adapts to user profile changes while at the same time being scalable and efficient in terms of consumption of the device resources.

Efficiency of human activity on information spreading on Twitter

In this work we study Twitter data to understand influence dynamics in social networks. We define user efficiency on Twitter, as the ratio between the emergent spreading process and the activity employed by the user. We characterize this property by means of a quantitative analysis of the structural and dynamical patterns emergent from human interactions, and show it to be universal across several Twitter conversations.

Human autoantibody recognition of DNA.

Combinatorial IgG Fab phage display libraries prepared from a systemic lupus erythematosus (SLE) donor and a healthy donor were affinity selected against human placental DNA. Human monoclonal antibody Fab fragments specific for DNA were isolated from both libraries, although Fabs of the highest affinity were isolated only from the lupus library. Generally, apparent affinities of the Fabs for human placental DNA, purified double-stranded DNA, and denatured DNA were approximately equivalent. Surface plasmon resonance indicated Fab binding constants for a double-stranded oligodeoxynucleotide of 0.2-1.3 x 10(8) M-1. The higher-affinity Fabs, as ranked by binding to human placental DNA or to the oligonucleotide probe, tested positive in the Crithidia luciliae assay commonly used in the diagnosis of SLE, and interestingly the genes encoding the heavy-chain variable regions of these antibodies displayed evidence of only minimal somatic hypermutation. The heavy chains of the SLE Fabs were characterized by a predominance of basic residues toward the N terminus of complementarity-determining region 3 (CDR3). The crucial role of heavy-chain CDR3 (HCDR3) in high-affinity DNA recognition was suggested by the creation of DNA binding in an unrelated antibody by HCDR3 transplantation from SLE antibodies. We propose that high-affinity DNA-binding antibodies can arise in SLE without extensive somatic hypermutation in the variable-region genes because of the expression of inappropriate HCDR3s.