Human Activity Recognition Based on Symbolic Representation Algorithms for Inertial Sensors

Abstract

Mobile sensing has allowed the emergence of a variety of solutions related to the monitoring and recognition of human activities (HAR). Such solutions have been implemented in smartphones for the purpose of better understanding human behavior. However, such solutions still suffer from the limitations of the computing resources found on smartphones. In this sense, the HAR area has focused on the development of solutions of low computational cost. In general, the strategies used in the solutions are based on shallow and deep learning algorithms. The problem is that not all of these strategies are feasible for implementation in smartphones due to the high computational cost required, mainly, by the steps of data preparation and the training of classification models. In this context, this article evaluates a new set of alternative strategies based on Symbolic Aggregate Approximation (SAX) and Symbolic Fourier Approximation (SFA) algorithms with the purpose of developing solutions with low computational cost in terms of memory and processing. In addition, this article also evaluates some classification algorithms adapted to manipulate symbolic data, such as SAX-VSM, BOSS, BOSS-VS and WEASEL. Experiments were performed on the UCI-HAR, SHOAIB and WISDM databases commonly used in the literature to validate HAR solutions based on smartphones. The results show that the symbolic representation algorithms are faster in the feature extraction phase, on average, by 84.81%, and reduce the consumption of memory space, on average, by 94.48%, and they have accuracy rates equivalent to conventional algorithms.

Reconhecimento de atividades humanas baseado na análise de fluxo contínuo de dados simbólicos

“…O caminho que proporcionou a concepção do NOHAR gerou várias publicações de impacto como [29], [28], [27], [9], [18] e [21]. Além disso, também tivemos uma publicação de capítulo de livro [4].…”

“…Assim, sempre que os dados brutos contidos na memória são processados, a memória é liberada para armazenar os novos dados coletados do fluxo contínuo de dados [7]. Paralelamente ao uso de algoritmos online, o Sousa Lima et al [28] propôs modelos de classificação offline de baixo custo baseados em algoritmos de representação simbólica. Esse tipo de algoritmo compacta os dados brutos em um conjunto de símbolos representados por um pacote de palavras.…”

“…A literatura apresenta dois algoritmos de representação simbólica que podem ser utilizados para discretizar os dados, o Symbolic Aggregate Approximation (SAX) [16] e o Symbolic Fourier Approximation (SFA) [23]. Sousa Lima et al [28] mostrou que esses dois algoritmos podem reduzir o consumo de memória em até 94,48%. Além disso, os algoritmos de classificação offline adaptados para lidar com dados simbólicos (por exemplo, SAX-VSM [24] e BOSS [22] são em média 6 vezes mais rápidos do que os algoritmos offline tradicionais no contexto HAR.…”

Systematic Analysis of a Military Wearable Device Based on a Multi-Level Fusion Framework: Research Directions

“…To the best of our knowledge, it is difficult for wearable sensors to recognize fast motion; however, it is a research issue that will inevitably arise in complex battlefield environments. The final issue is computing cost [105]. Compared with the other systems in the MLFF, the BTS has the largest number of sensors, which means that we need to conduct data mining on behavioral data with a high number of dimensions.…”

MBOSS: A Symbolic Representation of Human Activity Recognition Using Mobile Sensors

“…This section describes the symbolic representation algorithms used for time series as the basis for developing the method proposed in this work. Previous studies [20] have shown the potential to extract discrete features from inertial sensors signals represented by histograms based on word frequency distribution. The main idea of these algorithms is to reduce data dimensionality and noise of the time series in order to improve the classification stage of the machine learning algorithms.…”