Multimodal Emotion Recognition Using Bi-LG-GCN for MELD Dataset

Balkan Journal of Electrical and Computer Engineering, Journal Year: 2024, Volume and Issue: 12(1), P. 36 - 46

Published: March 1, 2024

Emotion recognition using multimodal data is a widely adopted approach due to its potential enhance human interactions and various applications. By leveraging for emotion recognition, the quality of can be significantly improved. We present Multimodal Lines Dataset (MELD) novel method bi-lateral gradient graph neural network (Bi-LG-GNN) feature extraction pre-processing. The dataset uses fine-grained labeling textual, audio, visual modalities. This work aims identify affective computing states successfully concealed in textual audio sentiment analysis. use pre-processing techniques improve consistency increase dataset’s usefulness. process also includes noise removal, normalization, linguistic processing deal with variances background discourse. Kernel Principal Component Analysis (K-PCA) employed extraction, aiming derive valuable attributes from each modality encode labels array values. propose Bi-LG-GCN-based architecture explicitly tailored effectively fusing Bi-LG-GCN system takes modality's feature-extracted pre-processed representation as input generator network, generating realistic synthetic samples that capture relationships. These generated samples, reflecting relationships, serve inputs discriminator which has been trained distinguish genuine data. With this approach, model learn discriminative features make accurate predictions regarding subsequent emotional states. Our was evaluated on MELD dataset, yielding notable results terms accuracy (80%), F1-score (81%), precision recall (81%) when dataset. steps discrimination. featuring synthesis, outperforms contemporary techniques, thus demonstrating practical utility.

Language: Английский

---

Automated Facial Emotion Recognition Using the Pelican Optimization Algorithm with a Deep Convolutional Neural Network

Electronics, Journal Year: 2023, Volume and Issue: 12(22), P. 4608 - 4608

Published: Nov. 11, 2023

Facial emotion recognition (FER) stands as a pivotal artificial intelligence (AI)-driven technology that exploits the capabilities of computer-vision techniques for decoding and comprehending emotional expressions displayed on human faces. With use machine-learning (ML) models, specifically deep neural networks (DNN), FER empowers automatic detection classification broad spectrum emotions, encompassing surprise, happiness, sadness, anger, more. Challenges in include handling variations lighting, poses, facial expressions, well ensuring model generalizes to various emotions populations. This study introduces an automated using pelican optimization algorithm with convolutional network (AFER-POADCNN) model. The primary objective AFER-POADCNN lies emotions. To accomplish this, median-filtering (MF) approach remove noise present it. Furthermore, capsule-network (CapsNet) can be applied feature-extraction process, allowing capture intricate nuances. optimize CapsNet model’s performance, hyperparameter tuning is undertaken aid (POA). ensures finely tuned detect wide array effectively across diverse populations scenarios. Finally, different kinds take place bidirectional long short-term memory (BiLSTM) network. simulation analysis system tested benchmark dataset. comparative result showed better performance over existing maximum accuracy 99.05%.

Language: Английский

---

Motion position prediction and machining accuracy compensation of galvanometer scanner based on BWO-GRU model

Mechanical Systems and Signal Processing, Journal Year: 2024, Volume and Issue: 210, P. 111081 - 111081

Published: Jan. 21, 2024

Language: Английский

---

Advancing Sentiment Analysis: A Novel LSTM Framework with Multi-head Attention

Published: April 30, 2025

This work proposes an LSTM-based sentiment classification model with multi-head attention mechanism and TF-IDF optimization. Through the integration of feature extraction attention, significantly improves text analysis performance. Experimental results on public data sets demonstrate that new method achieves substantial improvements in most critical metrics like accuracy, recall, F1-score compared to baseline models. Specifically, accuracy 80.28% test set, which is improved by about 12% comparison standard LSTM Ablation experiments also support necessity all modules, impact greatest performance improvement. research provides a proper approach analysis, can be utilized opinion monitoring, product recommendation, etc.

Language: Английский

---

Impact of social media on the evolution of English semantics through linguistic analysis

Forum for Linguistic Studies, Journal Year: 2024, Volume and Issue: 6(2)

Published: March 20, 2024

Social media (SM) influences social interaction in the age of digital media, impacting how languages develop. Since these networks play a role daily life, they create new words and conceptual frameworks that define our contemporary society. The current investigation investigates Twitter, Facebook, Reddit SM posts applying textual extraction. seven-year temporal sample demonstrates significant semantic change caused by society technology. analysis notices importance words, phrase meaning evolving, sentiment changes users' English usage, proving their adaptability. growing popularity phrases like eavesdropping doom-scrolling indicated life impact. This distinguishes each platform's unique linguistic features developments understanding language flow leading research future.

Language: Английский

---

HDEL: a hierarchical deep ensemble approach for text-based emotion detection

Multimedia Tools and Applications, Journal Year: 2024, Volume and Issue: unknown

Published: April 17, 2024

Language: Английский

---

Sentiment Analysis and Emotion Recognition in Social Media: A Comprehensive Survey

Applied Soft Computing, Journal Year: 2025, Volume and Issue: unknown, P. 112958 - 112958

Published: March 1, 2025

Language: Английский

---

nBERT: Harnessing NLP for Emotion Recognition in Psychotherapy to Transform Mental Health Care

Information, Journal Year: 2025, Volume and Issue: 16(4), P. 301 - 301

Published: April 9, 2025

The rising prevalence of mental health disorders, particularly depression, highlights the need for improved approaches in therapeutic interventions. Traditional psychotherapy relies on subjective assessments, which can vary across therapists and sessions, making it challenging to track emotional progression therapy effectiveness objectively. Leveraging advancements Natural Language Processing (NLP) domain-specific Large Models (LLMs), this study introduces nBERT, a fine-tuned Bidirectional Encoder Representations from Transformers (BERT) model integrated with NRC Emotion Lexicon, elevate emotion recognition transcripts. goal is provide computational framework that aids identifying patterns, tracking patient-therapist alignment, assessing outcomes. Addressing challenge classification text-based where non-verbal cues are absent, nBERT demonstrates its ability extract nuanced insights unstructured textual data, providing data-driven approach enhance assessments. Trained dataset 2021 transcripts, achieves an average precision 91.53%, significantly outperforming baseline models. This capability not only improves diagnostic accuracy but also supports customization strategies. By automating interpretation complex dynamics psychotherapy, exemplifies transformative potential NLP LLMs revolutionizing care. Beyond enables broader LLM applications life sciences, including personalized medicine healthcare.

Language: Английский

---

EEG-Based Emotion Classification in Financial Trading Using Deep Learning: Effects of Risk Control Measures

Sensors, Journal Year: 2023, Volume and Issue: 23(7), P. 3474 - 3474

Published: March 26, 2023

Day traders in the financial markets are under constant pressure to make rapid decisions and limit capital losses response fluctuating market prices. As such, their emotional state can greatly influence decision-making, leading suboptimal outcomes volatile conditions. Despite use of risk control measures such as stop loss orders, it is unclear if these strategies have a substantial impact on traders. In this paper, we aim determine orders has significant compared when not applied. The paper provides technical framework for valence-arousal classification trading using EEG data deep learning algorithms. We conducted two experiments: first experiment employed predetermined lock profit objectives, while second did employ or losses. also proposed novel hybrid neural architecture that integrates Conditional Random Field with CNN-BiLSTM model employs Bayesian Optimization systematically optimal hyperparameters. best obtained accuracies 85.65% 85.05% experiments, outperforming previous studies. Results indicate emotions associated Low Valence High Arousal, fear worry, were more prevalent experiment. hope, employing loss. contrast, Arousal (calmness) most prominent group which engage activities. Our results demonstrate efficacy our emotion aid risk-related decision-making abilities day Further, present limitations current work directions future research.

Language: Английский

---

An innovative EEG-based emotion recognition using a single channel-specific feature from the brain rhythm code method

Frontiers in Neuroscience, Journal Year: 2023, Volume and Issue: 17

Published: July 20, 2023

Introduction Efficiently recognizing emotions is a critical pursuit in brain–computer interface (BCI), as it has many applications for intelligent healthcare services. In this work, an innovative approach inspired by the genetic code bioinformatics, which utilizes brain rhythm features consisting of δ, θ, α, β, or γ, proposed electroencephalography (EEG)-based emotion recognition. Methods These are first extracted from sequencing technique. After evaluating them using four conventional machine learning classifiers, optimal channel-specific feature that produces highest accuracy each emotional case identified, so recognition through minimal data realized. By doing so, complexity can be significantly reduced, making more achievable practical hardware setups. Results The best classification accuracies achieved DEAP and MAHNOB datasets range 83–92%, SEED dataset, 78%. experimental results impressive, considering employed. Further investigation shows their representative channels primarily on frontal region, associated rhythmic characteristics typical multiple kinds. Additionally, individual differences found, varies with subjects. Discussion Compared to previous studies, work provides insights into designing portable devices, only one electrode appropriate generate satisfactory performances. Consequently, would advance understanding rhythms, offers solution classifying EEG signals diverse BCI applications, including

Language: Английский

---