A Comparative Study of Food Intake Detection Using Artificial Neural Network and Support Vector Machine

Abstract

In Machine Learning applications, the selection of the classification algorithm depends on the problem at hand. This paper provides a comparison of the performance of the Support Vector Machine (SVM) and the Artificial Neural Network (ANN) for food intake detection. A combination of time domain (TD) and frequency domain (FD) features, extracted from signals captured using a jaw motion sensor, were used to train both types of classifiers. Data were collected from 12 subjects in free-living for a period of 24-hrs under unrestricted conditions. ANN with a different number of hidden layer neurons and SVMs with different kernels were trained using a leave one out cross validation scheme. ANN achieved an average accuracy of 86.86 ± 6.5 % whereas SVM (with linear kernel) achieved an average classification accuracy of 81.93 ± 9.22 %. Data collected from an independent subject in a separate study were used to evaluate the performance of these classifiers in-terms of the number of meals detected per day resulting in an accuracy of 72.72% for ANN and 63.63% for SVM. The results suggest that ANN may perform better than SVM for this specific problem.

Automatic, wearable-based, in-field eating detection approaches for public health research: a scoping review

“…Self-report methods. Six of the studies had participants self-report all daily activities, including eating, via a log or diary 33,38,40,53,60,64 , while six studies had participants self-report just eating activity 37,39,51,54,58,63 . In one study, participants were asked to record their eating episodes with a smartphone front-facing video camera to obtain ground-truth eating 47 .…”

“…1). Six of these articles 40,49,52,56,57,59 reported on more than one in-field study that fit our eligibility criteria, so N = 40 studies (from 33 articles) is considered to be the final sample size for the review.…”

“…The majority of studies (N = 26 of 40) used multi-sensor systems (incorporating > 1 wearable sensor) to automatically detect eating activity 18,33,34,[37][38][39][40][41][44][45][46]48,49,[51][52][53][54][56][57][58] . On average, 2.10 wearable sensors (SD = 0.96) were used in the studies, with a range of 1-4 sensors ( Table 1).…”

“…Approximately 63% (N = 25) of the 40 studies utilized an accelerometer (device that determines acceleration) either by itself (N = 4) or incorporated into a sensor system (N = 21) to detect eating activity 18,[37][38][39][40][41]43,45,46,48,49,51,53,[56][57][58][59]62 (Table 2). The second most frequently utilized wearable sensor was a gyroscope (device that determines orientation) (N = 15) 33,38,39,46,48,49,51,53,[56][57][58] , followed by a microphone (N = 8) 34,35,47,52,54,60,61 , a piezoelectric sensor (N = 7) 18,40-42,44,45 , a RF transmitter and receiver (N = 6) 18,40,41,44,45 , and a smartwatch camera (N = 5) 56,57 (Table 2). EMG electrodes 36,63,…”

Wearable Food Intake Monitoring Technologies: A Comprehensive Review

“…For example, for the inertial approach, Dong et al [106] adopted the naive Bayes classifier with the naive assumption of the independence of features for the classification problem. Farooq et al [140] proposed the artificial neural network (ANN) classifier for the AIM system mentioned in Section 5.6 and compared it with the SVM. The ANN shows better performance in this study.…”