Carlos Eric Galván-Tejada

Magnetic Field Feature Extraction and Selection for Indoor Location Estimation

User indoor positioning has been under constant improvement especially with the availability of new sensors integrated into the modern mobile devices, which allows us to exploit not only infrastructures made for everyday use, such as WiFi, but also natural infrastructure, as is the case of natural magnetic field. In this paper we present an extension and improvement of our current indoor localization model based on the feature extraction of 46 magnetic field signal features. The extension adds a feature selection phase to our methodology, which is performed through Genetic Algorithm (GA) with the aim of optimizing the fitness of our current model. In addition, we present an evaluation of the final model in two different scenarios: home and office building. The results indicate that performing a feature selection process allows us to reduce the number of signal features of the model from 46 to 5 regardless the scenario and room location distribution. Further, we verified that reducing the number of features increases the probability of our estimator correctly detecting the users location (sensitivity) and its capacity to detect false positives (specificity) in both scenarios.

Evolution of Indoor Positioning Technologies: A Survey

Indoor positioning systems (IPS) use sensors and communication technologies to locate objects in indoor environments. IPS are attracting scientific and enterprise interest because there is a big market opportunity for applying these technologies. There are many previous surveys on indoor positioning systems; however, most of them lack a solid classification scheme that would structurally map a wide field such as IPS, or omit several key technologies or have a limited perspective; finally, surveys rapidly become obsolete in an area as dynamic as IPS. The goal of this paper is to provide a technological perspective of indoor positioning systems, comprising a wide range of technologies and approaches. Further, we classify the existing approaches in a structure in order to guide the review and discussion of the different approaches. Finally, we present a comparison of indoor positioning approaches and present the evolution and trends that we foresee.

Location Identification Using a Magnetic-field-based FFT Signature

Abstract User indoor positioning has been under constant improvement especially with the availability of new sensors integrated to the modern mobile devices. These sensory devices allow us to exploit not only infrastructures made for every day use, such as Wifi, but also natural infrastructure, as is the case of natural magnetic fields. In this work, we propose a novel approach that takes advantage of the benefits of using the magnetic sensor incorporated in most modern mobile devices, and the negligible variations of the Earths magnetic field to position an individual with high accuracy. Most importantly, the methodology proposed allows us to avoid the burden of having to collect magnetic information in different directions in order to construct an accurate magnetic map, showing an improvement on methods that require the individuals to construct bigger magnetic maps that contain redundant information such as magnitude in different directions.

Evaluation of Four Classifiers as Cost Function for Indoor Location Systems

Abstract In our previous research work, we proposed a methodology that uses magnetic-field and multivariate methods to estimate user location in an indoor environment. In this paper, we propose the use of this methodology to evaluate the performance of four different classification algorithms: Random Forest, Nearest Centroid, K Nearest Neighbors and Artificial Neural Networks; each classifier will be considered as a cost function of a genetic algorithm (GA) used in the feature selection process task of the methodology. The motivation to evaluate the algorithms of classification was that several ILSs use a classification algorithm in order to estimate the location of the user, but the classifiers performance vary from application to application. In order to evaluate the performance of each classification algorithm, the following issues were considered: (1) the time of the training phase to obtain the final classification algorithm; (2) the number of features needed for getting the model; (3) the type of the features from the final model; and (4) the sensitivity and specificity of the model. Our results indicate that Nearest centroid is the classfier algorithm that is best suited to be implemented in an end-user application given the obtained results on the evaluated criteria for the indoor location system (ILS).

Magnetic-Field Feature Extraction for Indoor Location Estimation

User indoor positioning has been under constant improvement especially with the availability of new sensors integrated into the modern mobile devices. These sensory devices allow us to exploit not only infrastructures made for every day use, such as Wi-Fi, but also natural infrastructure, as is the case of natural magnetic fields. From our experience working with mobile devices and Magnetic-Field based location systems, we identify some issues that should be addressed to improve the performance of a Magnetic-Field based system, such as a reduction of the data to be analyzed to estimate an individual location. In this paper we propose a feature extraction process that uses magnetic-field temporal and spectral features to acquire a classification model using the capabilities of mobile phones. Finally, we present a comparison against well known spectral classification algorithms with the aim to ensure the reliability of the feature extraction process.

Multi-view stacking for activity recognition with sound and accelerometer data

Abstract Many Ambient Intelligence (AmI) systems rely on automatic human activity recognition for getting crucial context information, so that they can provide personalized services based on the current users’ state. Activity recognition provides core functionality to many types of systems including: Ambient Assisted Living, fitness trackers, behavior monitoring, security, and so on. The advent of wearable devices along with their diverse set of embedded sensors opens new opportunities for ubiquitous context sensing. Recently, wearable devices such as smartphones and smart-watches have been used for activity recognition and monitoring. Most of the previous works use inertial sensors (accelerometers, gyroscopes) for activity recognition and combine them using an aggregation approach, i.e., extract features from each sensor and aggregate them to build the final classification model. This is not optimal since each sensor data source has its own statistical properties. In this work, we propose the use of a multi-view stacking method to fuse the data from heterogeneous types of sensors for activity recognition. Specifically, we used sound and accelerometer data collected with a smartphone and a wrist-band while performing home task activities. The proposed method is based on multi-view learning and stacked generalization, and consists of training a model for each of the sensor views and combining them with stacking. Our experimental results showed that the multi-view stacking method outperformed the aggregation approach in terms of accuracy, recall and specificity.

Magnetic-Field Feature Reduction for Indoor Location Estimation Applying Multivariate Models

In the context of a magnetic field-based indoor location system, this paper proposes a feature extraction process that uses magnetic-field temporal and spectral features in order to develop a classification model of indoor places, using only a magnetometer included in popular smartphones. We initially propose 46 features, 26 derived from the spectral evolution and 20 from the temporal one, chosen because of the statistical potential to summarize the behavior of the signal. Nevertheless, in order to simplify the classification model, a genetic algorithm approach, combined with forward selection and back elimination strategies was applied. Our results show that is possible to reduce the magnetic-field signal features from 46 to only 6 features, and estimating the users location with even better precision.

Multivariate Feature Selection of Image Descriptors Data for Breast Cancer with Computer-Assisted Diagnosis

Breast cancer is an important global health problem, and the most common type of cancer among women. Late diagnosis significantly decreases the survival rate of the patient; however, using mammography for early detection has been demonstrated to be a very important tool increasing the survival rate. The purpose of this paper is to obtain a multivariate model to classify benign and malignant tumor lesions using a computer-assisted diagnosis with a genetic algorithm in training and test datasets from mammography image features. A multivariate search was conducted to obtain predictive models with different approaches, in order to compare and validate results. The multivariate models were constructed using: Random Forest, Nearest centroid, and K-Nearest Neighbor (K-NN) strategies as cost function in a genetic algorithm applied to the features in the BCDR public databases. Results suggest that the two texture descriptor features obtained in the multivariate model have a similar or better prediction capability to classify the data outcome compared with the multivariate model composed of all the features, according to their fitness value. This model can help to reduce the workload of radiologists and present a second opinion in the classification of tumor lesions.

An Analysis of Audio Features to Develop a Human Activity Recognition Model Using Genetic Algorithms, Random Forests, and Neural Networks

This work presents a human activity recognition (HAR) model based on audio features. The use of sound as an information source for HAR models represents a challenge because sound wave analyses generate very large amounts of data. However, feature selection techniques may reduce the amount of data required to represent an audio signal sample. Some of the audio features that were analyzed include Mel-frequency cepstral coefficients (MFCC). Although MFCC are commonly used in voice and instrument recognition, their utility within HAR models is yet to be confirmed, and this work validates their usefulness. Additionally, statistical features were extracted from the audio samples to generate the proposed HAR model. The size of the information is necessary to conform a HAR model impact directly on the accuracy of the model. This problem also was tackled in the present work; our results indicate that we are capable of recognizing a human activity with an accuracy of 85% using the HAR model proposed. This means that minimum computational costs are needed, thus allowing portable devices to identify human activities using audio as an information source.

Feature Selection for Place Classification through Environmental Sounds

Abstract In this work, an environmental audio classification scheme is proposed using a Chi squared filter as a feature selection strategy. Using feature selection (FS), the original 62 features characteristic vector can be optimized, and it can be used for environmental sound classification. These features are obtained using statistical analysis and frequency domain analysis. As a result, we obtain a reduced feature vector composed of 15 features: 11 statistical and 4 of the frequency domain. Using this reduced vector, a 10 class classification was done, using Support Vector machines (SVM) as classification method, the accuracy is higher than 90%.