FORMICARY SWARM OPTIMIZED DEEP CNN FOR FACIAL EMOTION RECOGNITION FROM HUMAN FACIAL EXPRESSIONS DOI

Manisha Balkrishna Sutar,

Asha Ambhaikar

Biomedical Engineering Applications Basis and Communications, Journal Year: 2024, Volume and Issue: 36(05)

Published: July 17, 2024

Facial emotion recognition (FER) is a dominant research area that captures the biological facial features and matches data with existing databases to analyze individual’s emotional state. Numerous techniques have been formulated for attaining effective FER. However, occlusions, different head positions, deformed faces, motion blur under unrestricted settings, complicated backgrounds make it complex images. In this paper, formicary swarm optimization-based deep convolutional neural network (FSO-opt DCNN) model utilized detection which JAFFE RAVDESS expression datasets are used. DCNNs proficient built-in feature extraction strategies from images map various expressions corresponding states adopted addition, intensity, directional, edge patterns as well correlation extracted utilizing hybrid textual pattern, RESNET 101 VGG 16-based modules assist DCNN attain informative high-resolution Further, optimization (FSO) incorporated effectively tunes capture relationships between learned excel FER capability. Evaluating metrics, face using dataset achieves notable efficiencies during training percentage (TP) of 90%, values 97.51%, 95.48%, 99.55%, 97.48%, 96.47%, minimum loss 2.49%. Simultaneously, demonstrates robust metric 96.75%, 98.49%, 95.01%, 96.72%, 97.59%, 3.25%. Finally, obtained results reveal efficacy FSO-opt DCNN, particularly in tasks, outperforms models across datasets, showcasing its versatility potential analysis applications.

Language: Английский

Load Margin Assessment of Power Systems Using Physics-Informed Neural Network with Optimized Parameters DOI Creative Commons
Murilo E. C. Bento

Energies, Journal Year: 2024, Volume and Issue: 17(7), P. 1562 - 1562

Published: March 25, 2024

Challenges in the operation of power systems arise from several factors such as interconnection large systems, integration new energy sources and increase electrical demand. These challenges have required development fast reliable tools for evaluating systems. The load margin (LM) is an important index stability but traditional methods determining LM consist solving a set differential-algebraic equations whose information may not always be available. Data-Driven techniques Artificial Neural Networks were developed to calculate monitor LM, present unsatisfactory performance due difficulty generalization. Therefore, this article proposes design method Physics-Informed parameters will tuned by bio-inspired algorithms optimization model. Physical knowledge regarding incorporated into PINN training process. Case studies carried out discussed IEEE 68-bus system considering N-1 criterion disconnection transmission lines. results obtained proposed showed lower error values Root Mean Square Error (RMSE), (MSE) Absolute Percentage (MAPE) indices than Levenberg-Marquard method.

Language: Английский

Citations

7
Optimization of Sizing of Battery Energy Storage System for Residential Households by Load Forecasting with Artificial Intelligence (AI): Case of EV Charging Installation DOI Creative Commons

Nopphamat Promasa,

Ekawit Songkoh,

Siamrat Phonkaphon

et al.

Energies, Journal Year: 2025, Volume and Issue: 18(5), P. 1245 - 1245

Published: March 4, 2025

This paper presents the optimization sizing of a battery energy storage system for residential use from load forecasting using AI. The solar rooftop panel installation and charging systems electric vehicles are connected to low-voltage electrical Metropolitan Electricity Authority (MEA). daily electricity demand future used long short-term memory (LSTM) technique in order analyze appropriate size (BESS) residences. capacity is 5.5 kWp, which produces an average 28.78 kWh/day. minimum actual month 67.04 kWh, comprising base vehicles, can determine as 21.03 kWh. For this research, will be presented find BESS by considering over month, equal 102.67 17.84 When comparing values with forecast, it significantly reduce cost BESS.

Language: Английский

Citations

0
Predicting the fracture load of asphalt concrete under TPB test using POA-optimized machine learning methods DOI
Hongwei Ling, Xianglong Wang, Chuan Lv

et al.

Construction and Building Materials, Journal Year: 2025, Volume and Issue: 470, P. 140580 - 140580

Published: March 4, 2025

Language: Английский

Citations

0
Pattern synthesis of linear array antennas based on Chaos Triangular Pelican Optimization Algorithm DOI Creative Commons

A. Liu,

Yan Liu, Zhuo Chen

et al.

Scientific Reports, Journal Year: 2025, Volume and Issue: 15(1)

Published: April 8, 2025

Language: Английский

Citations

0
RS-Xception: A Lightweight Network for Facial Expression Recognition DOI Open Access
Liefa Liao,

Shouluan Wu,

Chao Song

et al.

Electronics, Journal Year: 2024, Volume and Issue: 13(16), P. 3217 - 3217

Published: Aug. 14, 2024

Facial expression recognition (FER) utilizes artificial intelligence for the detection and analysis of human faces, with significant applications across various scenarios. Our objective is to deploy facial emotion network on mobile devices extend its application diverse areas, including classroom effect monitoring, human–computer interaction, specialized training athletes (such as in figure skating rhythmic gymnastics), actor training. Recent studies have employed advanced deep learning models address this task, though these often encounter challenges like subpar performance an excessive number parameters that do not align requirements FER embedded devices. To tackle issue, we devised a lightweight structure named RS-Xception, which straightforward yet highly effective. Drawing strengths ResNet SENet, integrates elements from Xception architecture. been trained FER2013 datasets demonstrate superior efficiency compared conventional models. Furthermore, assessed model’s CK+, FER2013, Bigfer2013 datasets, achieving accuracy rates 97.13%, 69.02%, 72.06%, respectively. Evaluation complex RAF-DB dataset yielded rate 82.98%. The incorporation transfer notably enhanced accuracy, 75.38% dataset, underscoring significance our research. In conclusion, proposed model proves be viable solution precise sentiment estimation. future, may deployed research purposes.

Language: Английский

Citations

3
Using Learning Focal Point Algorithm to Classify Emotional Intelligence DOI Open Access
Abdelhak Sakhi,

Salah-Eddine Mansour,

Abderrahim Sekkaki

et al.

Journal of Robotics and Control (JRC), Journal Year: 2024, Volume and Issue: 5(1), P. 263 - 270

Published: Feb. 1, 2024

Recognizing the fundamental role of learners' emotions in educational process, this study aims to enhance experiences by incorporating emotional intelligence (EI) into teacher robots through artificial and image processing technologies. The primary hurdle addressed is inadequacy conventional methods, particularly convolutional neural networks (CNNs) with pooling layers, imbuing intelligence. To surmount challenge, research proposes an innovative solution—introducing a novel learning focal point (LFP) layer replace resulting significant enhancements accuracy other vital parameters. distinctive contribution lies creation application LFP algorithm, providing approach emotion classification for robots. results showcase algorithm's superior performance compared traditional CNN approaches. In conclusion, highlights transformative impact algorithm on models and, consequently, emotionally intelligent This contributes valuable insights convergence education, implications future advancements field.

Language: Английский

Citations

1
Hybrid Neural Network Approach for Tea Leaf Disease Detection Using Pelican and Mayfly Optimization Algorithms DOI Creative Commons

Saja Bilal Hafedh Al-Karawi,

Hakan Koyuncu

Jurnal Riset Informatika, Journal Year: 2024, Volume and Issue: 6(2), P. 119 - 130

Published: March 11, 2024

This study addresses the problem of plant diseases and difficulty detecting them, it presents a unique technique for automatic detection tea leaf by combining neural networks optimization techniques. Our research uses curated database photographs that includes healthy diseased specimens. The network (CNN) is trained fine-tuned using algorithms. To increase disease identification accuracy, we used hybrid novel algorithm called (POA-MA) which Pelican Optimization Algorithm (POA), Mayfly (MA) feature selection, followed classification with Support Vector Machine (SVM). suggested mechanism performance evaluated MSE, F-score, recall, sensitivity measures. CNN-POAMA model yielded 94.5%, 0.035, 0.91, 0.93, 0.92, respectively. advances precision agriculture establishing strong framework automated detection, allowing early intervention, eventually enhancing crop health.

Language: Английский

Citations

1
Performance analysis of deep unified model for facial expression recognition using convolution neural network DOI Open Access

Kavita Kavita,

Rajender Singh Chhillar

International Journal of Power Electronics and Drive Systems/International Journal of Electrical and Computer Engineering, Journal Year: 2024, Volume and Issue: 14(4), P. 4046 - 4046

Published: June 4, 2024

Facial expression recognition has gathered substantial attention in computer vision applications, with the need for robust and accurate models that can decipher human emotions from facial images. Performance analysis of a novel hybrid model combines strengths residual network (ResNet) dense (DenseNet) architectures after applying preprocessing recognition. The proposed capitalizes on complementary characteristics ResNet's DenseNet's densely connected blocks to enhance model's capacity extract discriminative features This research evaluates performance conducts comprehensive benchmark against established convolution neural (CNN) models. encompasses key aspects performance, including classification accuracy, adaptability LFW dataset expressions such as Anger, Fear, Happy, Disgust, Sad, Surprise, along Neutral. observes is more efficient computationally than existing consistently. eliminates information perspective further research.

Language: Английский

Citations

1
PH-CBAM: A Parallel Hybrid CBAM Network with Multi-Feature Extraction for Facial Expression Recognition DOI Open Access
Liefa Liao,

Shouluan Wu,

Chao Song

et al.

Electronics, Journal Year: 2024, Volume and Issue: 13(16), P. 3149 - 3149

Published: Aug. 9, 2024

Convolutional neural networks have made significant progress in human Facial Expression Recognition (FER). However, they still face challenges effectively focusing on and extracting facial features. Recent research has turned to attention mechanisms address this issue, primarily local feature details rather than overall Building upon the classical Block Attention Module (CBAM), paper introduces a novel Parallel Hybrid Model, termed PH-CBAM. This model employs split-channel enhance extraction of key features while maintaining minimal parameter count. The proposed enables network emphasize relevant during expression classification. Heatmap analysis demonstrates that PH-CBAM highlights information. By employing multimodal approach initial image phase, structure captures various algorithm integrates residual MISH activation function create multi-feature network, addressing issues such as gradient vanishing negative zero point transmission. enhances retention valuable information facilitates flow between target images. Evaluation benchmark datasets FER2013, CK+, Bigfer2013 yielded accuracies 68.82%, 97.13%, 72.31%, respectively. Comparison with mainstream models FER2013 CK+ efficiency model, comparable accuracy current advanced models, showcasing its effectiveness emotion detection.

Language: Английский

Citations

1
Facial recognition and classification for customer information systems: a feature fusion deep learning approach with FFDMLC algorithm DOI

M. Prithi,

K. Tamizharasi

Computing, Journal Year: 2024, Volume and Issue: unknown

Published: Oct. 4, 2024

Language: Английский

Citations

1