Facial Expression recognition via neurons partially activated discriminated ELM

Wang, Lucai; Liu, Mao; Lu, Xiao; Lin, Haijun

doi:10.1109/wcica.2018.8630457

Cited by 1 publication

(6 citation statements)

References 5 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…A novel contribution was made by combining back-propagation neural network (BPNN), discrete cosine transform (DCT) and histograms of oriented gradient (HOG) to complete the task of face classification via feature extraction [47]. SVMs were employed to train the data using Gabor wavelets and classification in [48]. Additionally, a unique approach to facial recognition was developed based on an extreme learning machine (ELM) to impose the l-norm on the hidden weight matrix and identify the active neurons, resulting in a simple network topology [48].…”

Section: Face-basedmentioning

confidence: 99%

“…SVMs were employed to train the data using Gabor wavelets and classification in [48]. Additionally, a unique approach to facial recognition was developed based on an extreme learning machine (ELM) to impose the l-norm on the hidden weight matrix and identify the active neurons, resulting in a simple network topology [48].…”

Section: Face-basedmentioning

confidence: 99%

“…[21] Unknown Primary The data were acquired from 16 study participants using the microphones of the Pepper robot for audio recording, and the robot captured the images of the user drawings for the visuospatial and executive skills, which were then analysed. [48], [45], [33], [24], [50], [38], [53], [54] The Japanese Female Facial Expression (JAFFE) Dataset Secondary It features 213 facial photos of 10 Japanese women, each with seven different facial expressions: anger, disgust, fear, happiness, neutral, sadness and surprise. The image has a resolution of 256 × 256 pixels.…”

Section: Primarymentioning

confidence: 99%

“…In addition, the primary challenge is finding a classifier with high generalisation performance [48] [26] [52]. One of the reasons may be that the standard SoftMax loss function has limited discriminative ability [42].…”

Section: Challenges Related To Classification Accuracymentioning

confidence: 99%

“…Furthermore, the number of neurons is hard to determine in a given problem by utilising the traditional ELM classifier. Improving recognition accuracy whilst using few neurons and obtaining more sparse solutions than those provided by the typical ELM method is necessary [48].…”

Section: Challenges Related To Classification Accuracymentioning

confidence: 99%

See 4 more Smart Citations

Robotics Utilization in Automatic Vision-Based Assessment Systems From Artificial Intelligence Perspective: A Systematic Review

et al. 2022

View full text Add to dashboard Cite

This study presents a systematic review of the automatic vision-based assessment systems and models from related fields over the last 6 years. Many studies focus on automatic vision-based assessment in the area of robotics, artificial intelligence and virtual reality or augmented reality (AR) to enhance the communication between humans and machines for general purposes. Three reliable databases, IEEEXplore, Science Direct and Web of Science, were used to obtain relevant studies on the given topic. Several stages of filtering and scanning were applied according to the inclusion/exclusion criteria to filter the obtained 3505 papers from 2015 to 2020. Finally, 97 papers met the criteria. They were classified into four main categories based on their field by following the scientific taxonomy: computer-vision-based, robotics-based, AR-based and hybrid-based categories, accounting for 42.26% (n = 41/97 papers), 48.45% (n = 47/97 papers), 6.18% (n = 6/97 papers) and 4.12% (n = 4/97 papers), respectively. Subsequently, a deep and critical analysis of this multifield systematic review highlighted new research opportunities, motivations, challenges and recommendations that need attention to integrate interdisciplinary studies. Thus, automatic vision-based assessment, which is a field requiring automated solutions, tools and methods, enhances the ability of the assistive technology and facilitates the interaction of individuals with machines. Many studies have been conducted on the automatic vision-based assessment systems and their subtypes to promote accurate communication and performance evaluation in human-machine interaction. This study can provide researchers with useful guides and valuable information for future research. This study also addresses the ambiguity of automatic vison-based assessment models in interdisciplinary fields.

show abstract

Section: Face-basedmentioning

confidence: 99%

Section: Face-basedmentioning

confidence: 99%