Multiclass Least-Square Support Vector Machine for Myoelectric-Based Facial Gesture Recognition

Hamedi, Mahyar; Salleh, Shaharuddin; Noor, Alias Mohd; Harris, Arief R.; Majid, Norazman Abdul

doi:10.1007/978-3-319-02913-9_46

Cited by 3 publications

(5 citation statements)

References 13 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Compared to [39], classification accuracy is improved by 2.2%, and processing time is reduced about 0.6 second which again emphasizes the usefulness of EMG denoising rather than just using multi-feature sets. It must be highlighted that in this study the number of classes is also increased to eleven which affects the system performance.…”

Section: Discussionmentioning

confidence: 73%

“…According to Table 1, as far as the maximum number of classes (10) is concerned, up to 97.1 % classification accuracy was achieved where Least Square Support Vector Machine (LS-SVM) algorithm classified the fused feature set including Mean Absolute Value (MAV) and RMS [39]. However, the reported 1.37 seconds processing time seems not to be suitable for real-time applications.…”

Section: Introductionmentioning

confidence: 84%

“…In this method, equality constraints are used to find the optimal solution by solving a set of linear equations instead of solving a quadratic optimization problem [81]. This classifier has already been examined for facial EMG classification and promising results were reported in [34], [39] and [42]. The LS-SVM model used in this paper is formed using RBF kernel where the regularization    and smoothing    2 parameters are set at 10 and 0.2 respectively while the multiclass LS-SVM is trained and encoded by the one-versus-all scheme.…”

Section: Least-square Support Vector Machines (Ls-svms)mentioning

confidence: 99%

“…However, in previous studies this behavior has not been widely considered since only very few methods for either feature extraction or classification have been investigated. Classification accuracy has always been the most popular measure to assess the system performance; however, computational cost as another essential metric has only been considered in [29], [36][37][38][39] and [42].…”

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations

Robust Facial Expression Recognition for MuCI: A Comprehensive Neuromuscular Signal Analysis

Hamedi

Salleh

Ting

et al. 2018

IEEE Trans. Affective Comput.

Self Cite

View full text Add to dashboard Cite

This paper presents a comprehensive study on the analysis of neuromuscular signal activities to recognize eleven facial expressions for Muscle Computer Interfacing applications. A robust denoising protocol comprised of Wavelet transform and Kalman filtering is proposed to enhance the electromyogram (EMG) signal-to-noise ratio and improve classification performance. The effectiveness of eight different time-domain facial EMG features on system performance is examined and compared in order to identify the most discriminative one. Fourteen pattern recognition-based algorithms are employed to classify the extracted features. These classifiers are evaluated in terms of classification accuracy and processing time. Finally, the best methods that obtain almost identical system performance are compared through the Normalized Mutual Information (NMI) criterion and a repeated measure analysis of variance (ANOVA) for a statistical significant test.To clarify the impact of signal denoising, all considered EMG features and classifiers are assessed with and without this stage. Results show that: (1) the proposed denosing step significantly improves the system performance; (2) Root Mean Square is the most discriminative facial EMG feature; (3) discriminant analysis when the parameters are estimated by the Maximum Likelihood algorithm achieves the highest classification accuracy and NMI; however, ANOVA reveals no significant difference among the best methods with almost similar performance.

show abstract

Section: Discussionmentioning

confidence: 73%

Section: Introductionmentioning

confidence: 84%

Section: Least-square Support Vector Machines (Ls-svms)mentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Robust Facial Expression Recognition for MuCI: A Comprehensive Neuromuscular Signal Analysis

Hamedi

Salleh

Ting

et al. 2018

IEEE Trans. Affective Comput.

Self Cite

View full text Add to dashboard Cite

show abstract

“…Their major advantage is that they are fast to calculate as no mathematical transformation is needed. Recently, we focused on analysing facial (SEMG) temporal characteristics for facial gesture recognition [3][4][5][6][7][8][9][10]. In these studies, various types of time-domain features were evaluated to find the one that provided the most discriminating characteristic for recognizing different facial gestures.…”

Section: Introductionmentioning

confidence: 99%

Time-frequency facial gestures EMG analysis using bilinear distribution

Hamedi

Salleh

Ismail

et al. 2015

2015 IEEE International Conference on Signal and Image Processing Applications (ICSIPA)

Self Cite

View full text Add to dashboard Cite

Electromyogram (EMG)-based facial gesture recognition has recently drawn the researchers' attention as a potential medium in different areas, particularly in assistive technology and rehabilitation. Efficient analysis of facial neuromuscular signals generated by different facial muscles can provide lots of information about underlying facial movement mechanisms which can be used to characterize different facial gestures as well as muscle abnormalities like in patients with muscular dystrophy or facial palsy. This paper investigated timevarying properties of facial EMGs in time-frequency domain. Significantly we studied changes of EMG spectrum across time while performing ten different facial gestures. The facial gestures were recorded from ten individuals through three bipolar pairs of surface electrodes. Time-Frequency analysis was carried out using B-Distribution to resolve EMG components in timefrequency domain and specify the signal frequency components that change over time. We observed that 1) there were no significant differences among facial gestures EMG time-varying spectrum distributions, 2) EMG power spectrum decreased over time in each epoch after about one second from the beginning of each movement, 3) the most significant power spectrum of facial EMGs was within 60-300 Hz.

show abstract