Abstract
Human activity recognition (HAR) has become an increasingly popular application of machine learning across a range of domains. Typically the HAR task that a machine learning algorithm is trained for requires separating multiple activities such as walking, running, sitting, and falling from each other. Despite a large body of work on multi-class HAR, and the well-known fact that the performance on a multi-class problem can be significantly affected by how it is decomposed into a set of binary problems, there has been little research into how the choice of multi-class decomposition method affects the performance of HAR systems. This paper presents the first empirical comparison of multi-class decomposition methods in a HAR context by estimating the performance of five machine learning algorithms when used in their multi-class formulation, with four popular multi-class decomposition methods, five expert hierarchies—nested dichotomies constructed from domain knowledge—or an ensemble of expert hierarchies on a 17-class HAR data-set which consists of features extracted from tri-axial accelerometer and gyroscope signals. We further compare performance on two binary classification problems, each based on the topmost dichotomy of an expert hierarchy. The results show that expert hierarchies can indeed compete with one-vs-all, both on the original multi-class problem and on a more general binary classification problem, such as that induced by an expert hierarchy’s topmost dichotomy. Finally, we show that an ensemble of expert hierarchies performs better than one-vs-all and comparably to one-vs-one, despite being of lower time and space complexity, on the multi-class problem, and outperforms all other multi-class decomposition methods on the two dichotomous problems.
Highlights
Human activity recognition (HAR) systems have become an increasingly popular application area for machine learning across a range of domains, including the medical, industrial, home care and assisted livingdomains
We make four main contributions: (1) the first empirical evaluation of the effect that the choice of multi-class decomposition method has on the performance of various binary learning algorithms on a multi-class HAR problem; (2) the first direct comparison of hierarchical classification that is guided by domain knowledge with standard domain-agnostic multi-class decomposition methods on a multi-class HAR problem; (3) we formulate a threshold that indicates when a nested dichotomy’s branch cannot possibly be on the path to the predicted class, and does not need to be explored; and (4) we show that domain knowledge can be used to construct a multi-class classifier that has lower computational complexity and is easier to interpret than, but performs comparable to one-vs-all
Our analysis shows that the ensemble of gradient boosted trees significantly and consistently outperforms the other algorithms, both on the original 17-class problem and on the two dichotomous problems induced by the topmost dichotomy of EH1 and EH4
Summary
Human activity recognition (HAR) systems have become an increasingly popular application area for machine learning across a range of domains, including the medical (e.g., monitoring ambulatory patients [1]), industrial (e.g., monitoring workers for movements with increased risk of repetitive strain injury [2]), home care and assisted living (e.g., monitoring the elderly for dangerous falls [3])domains. To apply machine learning classification algorithms that are designed to deal with two target classes, such a multi-class problem must first be decomposed into a set of binary problems. A separate instance of the machine learning algorithm is trained for each of these binary problems. Besides making the multi-class problem amenable to binary classification algorithms, there are other benefits to decomposing a multi-class problem into a set of dichotomous problems. Perhaps the most important benefit is that a binary classification algorithm can be subjected to any of a number of well-known analyses, such as Receiver Operator Characteristic (ROC) or Sensitivity/Specificity analysis, which can yield insights that serve to tune the classifier with respect to the relative cost of false positives and false negatives
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callDisclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
