An Innovative Multi-Model Neural Network Approach for Feature Selection in Emotion Recognition Using Deep Feature Clustering

Asghar, Muhammad Adeel; Khan, Muhammad Jamil; Rizwan, Muhammad; Mehmood, Raja Majid; Kim, Sun‐Hee

doi:10.3390/s20133765

Cited by 29 publications

(19 citation statements)

References 51 publications

(65 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The majority of these models were mostly based on a single modality, such as EEG [22], [24], [25] and RESP [23] without feature fusion. Other models used one type of late fusion that was applied on multi deep models using a single modality [21] with deep feature clustering [26], or multi modalities [14], but other types of fusion were not explored. In [14], only one combination of RESP and Heart Rate Variability (HRV) with late fusion was implemented.…”

Section: Related Work and Objectivesmentioning

confidence: 99%

Multi-Modal Physiological Data Fusion for Affect Estimation Using Deep Learning

Hssayeni

Ghoraani

2021

IEEE Access

View full text Add to dashboard Cite

Automated momentary estimation of positive and negative affects (PA and NA), the basic sense of feeling, can play an essential role in detecting the early signs of mood disorders. Physiological wearable sensors and machine learning have a potential to make such automated and continuous measurements. However, the physiological signals' features that are associated with the subject-reported PA or NA may not be known. In this work, we use data-driven feature extraction based on deep learning to investigate the application of raw physiological signals for estimating PA and NA. Specifically, we propose two multimodal data fusion methods with deep Convolutional Neural Networks. We use the proposed architecture to estimate PA and NA and also classify baseline, stress, and amusement emotions. The training and evaluation of the methods are performed using four physiological and one chest motion signal modalities collected using a chest sensing unit from 15 subjects. Overall, our proposed model performed better than traditional machine learning on hand-crafted features. Utilizing only two modalities, our proposed model estimated PA with a correlation of 0.69 (p<0.05) vs. 0.59 (p<0.05) with traditional machine learning. These correlations were 0.79 (p<0.05) vs. 0.73 (p<0.05) for NA estimation. The best emotion classification was achieved by the traditional method with 79% F1-score and 80% accuracy when all the four physiological modalities are used. This is while with only two modalities, the deep learning achieved 78% F1-score and 79% accuracy.

show abstract

Section: Related Work and Objectivesmentioning

confidence: 99%

Multi-Modal Physiological Data Fusion for Affect Estimation Using Deep Learning

Hssayeni

Ghoraani

2021

IEEE Access

View full text Add to dashboard Cite

show abstract

“…EMD is an adaptive method that derives fundamental functions directly from the data [ 36 ]. EMD does not require any previously known value of the signal for its computation.…”

Section: Methodsmentioning

confidence: 99%

Expert Hypertension Detection System Featuring Pulse Plethysmograph Signals and Hybrid Feature Selection and Reduction Scheme

Khan

Aziz

Akram

et al. 2021

Sensors

View full text Add to dashboard Cite

Hypertension is an antecedent to cardiac disorders. According to the World Health Organization (WHO), the number of people affected with hypertension will reach around 1.56 billion by 2025. Early detection of hypertension is imperative to prevent the complications caused by cardiac abnormalities. Hypertension usually possesses no apparent detectable symptoms; hence, the control rate is significantly low. Computer-aided diagnosis based on machine learning and signal analysis has recently been applied to identify biomarkers for the accurate prediction of hypertension. This research proposes a new expert hypertension detection system (EHDS) from pulse plethysmograph (PuPG) signals for the categorization of normal and hypertension. The PuPG signal data set, including rich information of cardiac activity, was acquired from healthy and hypertensive subjects. The raw PuPG signals were preprocessed through empirical mode decomposition (EMD) by decomposing a signal into its constituent components. A combination of multi-domain features was extracted from the preprocessed PuPG signal. The features exhibiting high discriminative characteristics were selected and reduced through a proposed hybrid feature selection and reduction (HFSR) scheme. Selected features were subjected to various classification methods in a comparative fashion in which the best performance of 99.4% accuracy, 99.6% sensitivity, and 99.2% specificity was achieved through weighted k-nearest neighbor (KNN-W). The performance of the proposed EHDS was thoroughly assessed by tenfold cross-validation. The proposed EHDS achieved better detection performance in comparison to other electrocardiogram (ECG) and photoplethysmograph (PPG)-based methods.

show abstract

“…For example, to complete different tasks effectively, different DNN models need to be trained, tuning the parameters through repeating the trial and error, which optimizes the model structure [29]. erefore, training a DNN model to effectively process specific tasks takes up days or even weeks of the entire computing cluster time [30]. In addition, the parameter optimization of the DNN model not only requires high-performance GPU, TPU, and other higher computer hardware environments but also has high requirements for datasets.…”

Section: Introductionmentioning

confidence: 99%

A Stochastic Learning Algorithm for Machine Fault Diagnosis

Dong

Liu

Kang

et al. 2022

Shock and Vibration

View full text Add to dashboard Cite

Industrial big data bring a large number of high-dimensional sample datasets. Although a deep learning network can well mine the internal nonlinear structure of the dataset, the construction of the deep learning model requires a lot of computing time and hardware facilities. At the same time, there are some nonlinear problems such as noise and fluctuation in industrial data, which make the deep architecture extremely complex and the recognition accuracy of the diagnosis model difficult to guarantee. To solve this problem, a new method, named stochastic learning algorithm (SL), is proposed in this paper for dimension reduction. The proposed method consists of three steps: firstly, to increase the computational efficiency of the model, the dimension of the high-dimensional data is reduced by establishing a random matrix; secondly, for enhancing the clustering influence of the sample, the input data are enhanced by feature processing; thirdly, to make the clustering effects more pronounced, the noise and interference of the data need to be processed, and the singularity value denoising method is used to denoise training data and test data. To further prove the superiority of the SL method, we conducted two sets of experiments on the wind turbine gearbox and the benchmark dataset. It can be seen from the experimental results that the SL method not only improves the classification accuracy but also reduces the computational burden.

show abstract

An Innovative Multi-Model Neural Network Approach for Feature Selection in Emotion Recognition Using Deep Feature Clustering

Cited by 29 publications

References 51 publications

Multi-Modal Physiological Data Fusion for Affect Estimation Using Deep Learning

Multi-Modal Physiological Data Fusion for Affect Estimation Using Deep Learning

Expert Hypertension Detection System Featuring Pulse Plethysmograph Signals and Hybrid Feature Selection and Reduction Scheme

A Stochastic Learning Algorithm for Machine Fault Diagnosis

Contact Info

Product

Resources

About