A Comparative Study of Feature Selection Approaches for Human Activity Recognition Using Multimodal Sensory Data

Abstract

Human activity recognition (HAR) aims to recognize the actions of the human body through a series of observations and environmental conditions. The analysis of human activities has drawn the attention of the research community in the last two decades due to its widespread applications, diverse nature of activities, and recording infrastructure. Lately, one of the most challenging applications in this framework is to recognize the human body actions using unobtrusive wearable motion sensors. Since the human activities of daily life (e.g., cooking, eating) comprises several repetitive and circumstantial short sequences of actions (e.g., moving arm), it is quite difficult to directly use the sensory data for recognition because the multiple sequences of the same activity data may have large diversity. However, a similarity can be observed in the temporal occurrence of the atomic actions. Therefore, this paper presents a two-level hierarchical method to recognize human activities using a set of wearable sensors. In the first step, the atomic activities are detected from the original sensory data, and their recognition scores are obtained. Secondly, the composite activities are recognized using the scores of atomic actions. We propose two different methods of feature extraction from atomic scores to recognize the composite activities, and they include handcrafted features and the features obtained using the subspace pooling technique. The proposed method is evaluated on the large publicly available CogAge dataset, which contains the instances of both atomic and composite activities. The data is recorded using three unobtrusive wearable devices: smartphone, smartwatch, and smart glasses. We also investigated the performance evaluation of different classification algorithms to recognize the composite activities. The proposed method achieved 79% and 62.8% average recognition accuracies using the handcrafted features and the features obtained using subspace pooling technique, respectively. The recognition results of the proposed technique and their comparison with the existing state-of-the-art techniques confirm its effectiveness.

SenseHunger: Machine Learning Approach to Hunger Detection Using Wearable Sensors

“…Hand-crafted Features: We used 18 hand-crafted features [ 9 , 35 ] consisting of the statistical and frequency-related values of the input signals. These features are listed in Table 2 .…”

“…The top decision node in a tree points to the best predictor, which is called the root node [ 47 ]. RF: This is a popular ensemble learning method used for various types of classification problems such as activity recognition [ 35 ], where multiple DTs are created at training time [ 48 , 49 , 50 , 51 , 52 ]. In RF, each tree casts a unit vote by assigning each input to the most likely class label.…”

Assessing Human Motion During Exercise Using Machine Learning: A Literature Review

“…Once all features are extracted and/or engineered, it is possible to have many features that will increase the complexity and accuracy of the ML model [41]. Therefore, to reduce complexity, a subset of features may be selected instead of using all of them to develop the model.…”

The Symphony of Team Flow in Virtual Teams. Using Artificial Intelligence for Its Recognition and Promotion

“…Therefore, based on the indicators described in Tables 4 , 5 , it is possible to develop an application for team flow analysis in the form of an end-to-end machine learning-based algorithm that takes input from multiple sensors including wearable devices, cameras and microphones, and predict the cognitive states of the participants of virtual teams by analyzing not only their own physiological data but also their interaction and communication with other team members. In order to build such an application, it is necessary to answer several questions like how to handle the heteroscedasticity of different input signals (i.e., the variability of variance of errors of the input data), what will be the useful features ( Amjad et al, 2021 ), which feature extraction and classifications techniques will be used? The artificial neural networks handle all these issues and are able to learn and model non-linear and complex relationships between input and output ( Li et al, 2018 , 2020 ).…”