Data Mining Practical Machine Learning Tools and Techniques

Deep learning for sensor-based activity recognition: A survey

An increase in world population along with a significant aging portion is forcing rapid rises in healthcare costs. The healthcare system is going through a transformation in which continuous monitoring of inhabitants is possible even without hospitalization. The advancement of sensing technologies, embedded systems, wireless communication technologies, nano technologies, and miniaturization makes it possible to develop smart systems to monitor activities of human beings continuously. Wearable sensors detect abnormal and/or unforeseen situations by monitoring physiological parameters along with other symptoms. Therefore, necessary help can be provided in times of dire need. This paper reviews the latest reported systems on activity monitoring of humans based on wearable sensors and issues to be addressed to tackle the challenges.

Wearable Sensors for Human Activity Monitoring: A Review

This paper proposes a deep convolutional neural network for HAR using smartphone sensors.Experiments show that the proposed method derives relevant and more complex features.The method achieved an almost perfect classification on moving activities.It outperforms other state-of-the-art data mining techniques in HAR. Human activities are inherently translation invariant and hierarchical. Human activity recognition (HAR), a field that has garnered a lot of attention in recent years due to its high demand in various application domains, makes use of time-series sensor data to infer activities. In this paper, a deep convolutional neural network (convnet) is proposed to perform efficient and effective HAR using smartphone sensors by exploiting the inherent characteristics of activities and 1D time-series signals, at the same time providing a way to automatically and data-adaptively extract robust features from raw data. Experiments show that convnets indeed derive relevant and more complex features with every additional layer, although difference of feature complexity level decreases with every additional layer. A wider time span of temporal local correlation can be exploited (1?9-1?14) and a low pooling size (1?2-1?3) is shown to be beneficial. Convnets also achieved an almost perfect classification on moving activities, especially very similar ones which were previously perceived to be very difficult to classify. Lastly, convnets outperform other state-of-the-art data mining techniques in HAR for the benchmark dataset collected from 30 volunteer subjects, achieving an overall performance of 94.79% on the test set with raw sensor data, and 95.75% with additional information of temporal fast Fourier transform of the HAR data set.

Human activity recognition with smartphone sensors using deep learning neural networks

Interest in Wireless Body Area Networks (WBANs) has increased significantly in recent years thanks to the advances in microelectronics and wireless communications. Owing to the very stringent application requirements in terms of reliability, energy efficiency, and low device complexity, the design of these networks requires the definition of new protocols with respect to those used in general purpose wireless sensor networks. This motivates the effort in research activities and in standardisation process of the last years. This survey paper aims at reporting an overview of WBAN main applications, technologies and standards, issues in WBANs design, and evolutions. Some case studies are reported, based on both real implementation and experimentation on the field, and on simulations. These results have the aim of providing useful insights for WBAN designers and of highlighting the main issues affecting the performance of these kind of networks.

A Survey on Wireless Body Area Networks: Technologies and Design Challenges

Providing accurate and opportune information on people's activities and behaviors is one of the most important tasks in pervasive computing. Innumerable applications can be visualized, for instance, in medical, security, entertainment, and tactical scenarios. Despite human activity recognition (HAR) being an active field for more than a decade, there are still key aspects that, if addressed, would constitute a significant turn in the way people interact with mobile devices. This paper surveys the state of the art in HAR based on wearable sensors. A general architecture is first presented along with a description of the main components of any HAR system. We also propose a two-level taxonomy in accordance to the learning approach (either supervised or semi-supervised) and the response time (either offline or online). Then, the principal issues and challenges are discussed, as well as the main solutions to each one of them. Twenty eight systems are qualitatively evaluated in terms of recognition performance, energy consumption, obtrusiveness, and flexibility, among others. Finally, we present some open problems and ideas that, due to their high relevance, should be addressed in future research.

A Survey on Human Activity Recognition using Wearable Sensors

Context-aware applications have been the focus of extensive research, yet their implementation in mobile devices usually becomes challenging due to restrictions of processing and energy. In this work, we present, demonstrate, and evaluate a mobile platform for real-time human activity recognition. The evaluation results indicate that the system can be implemented in real scenarios meeting accuracy, response time, and energy consumption requirements.

A mobile human activity recognition system

This paper presents ABLA, a novel multiobjective ant colony-based optimization algorithm to address the bin packing problem with load balancing. ABLA incorporates (1) a new probabilistic decision rule that builds solutions by making use of individual pheromone matrices for each objective function; (2) a new pheromone updating approach in which ants deposit variable amounts of pheromone; (3) two new local search methods to improve load balancing: LBH and LBH-AB; and (4) the Pareto dominance approach to select optimal solutions. ABLA is compared to a Multiobjective Max-Min Ant System (MO-MMAS) and an adapted multiobjective version of the First-Fit Decreasing (FFD) algorithm, which is the best known ρ-approximation algorithm for the bin packing problem. Results show that ABLA finds better solutions than both FFD and MO-MMAS, and that LBH and LBH-AB noticeably improve the load balancing across bins.

A multiobjective ant colony-based optimization algorithm for the bin packing problem with load balancing

The identification of psychological stress can provide important feedback in order to perform critical activities. While a certain amount of stress may increase performance, an overly stressful reaction may hinder it. Because subjective bias can make it difficult to accurately recognize psychological stress, it would be advantageous for an external system to perform the task instead. We present a platform for psychological stress detection using physiological sensors during a chess match. The sensors are inside an unobtrusive chest strap that can be worn by the player during a match. By playing games on an Android phone, the system can apply machine learning techniques to the player's vital sign data to give important feedback such as which moves caused the player to become stressed during a match.

Oscar D. Lara

Papers

A Survey on Human Activity Recognition using Wearable Sensors

A mobile human activity recognition system

A multiobjective ant colony-based optimization algorithm for the bin packing problem with load balancing

Recognizing mental stress in chess players using vital sign data