Attention-Based PSO-LSTM for Emotion Estimation Using EEG DOI Creative Commons

Hayato OKA,

Keiko Ono, Panagiotis Adamidis

и другие.

Sensors, Год журнала: 2024, Номер 24(24), С. 8174 - 8174

Опубликована: Дек. 21, 2024

Recent advances in emotion recognition through Artificial Intelligence (AI) have demonstrated potential applications various fields (e.g., healthcare, advertising, and driving technology), with electroencephalogram (EEG)-based approaches demonstrating superior accuracy compared to facial or vocal methods due their resistance intentional manipulation. This study presents a novel approach enhance EEG-based estimation by emphasizing temporal features efficient parameter space exploration. We propose model combining Long Short-Term Memory (LSTM) an attention mechanism highlight EEG data while optimizing LSTM parameters Particle Swarm Optimization (PSO). The assigned weights hidden states, PSO dynamically optimizes the vital parameters, including units, batch size, dropout rate. Using DEAP SEED datasets, which serve as benchmark datasets for research using EEG, we evaluate model’s performance. For dataset, conduct four-class classification of combinations high low valence arousal states. perform three-class negative, neutral, positive emotions dataset. proposed achieves 0.9409 on surpassing previous state-of-the-art 0.9100 reported Lin et al. attains 0.9732 recording one highest accuracies among related research. These results demonstrate that integrating significantly improves estimation, contributing advancement technology.

Язык: Английский

A Study of Ethical and Cultural Considerations in the Integration of Artificial Intelligence and the Art of Dance DOI Open Access

Yanze Sun,

Lin Chu

Applied Mathematics and Nonlinear Sciences, Год журнала: 2025, Номер 10(1)

Опубликована: Янв. 1, 2025

Abstract The involvement of artificial intelligence in the creation dance art is a phenomenon that cannot be ignored current field art. deeper dance, more unclear boundaries resulting ethical issues such as impact individual emotions, social trust, employment, blurring subject and whether ends. In order to address these issues, this paper adopts questionnaire survey method investigate public’s views on integration art, explores cultural considerations from what perspective. number people who think development requires certain constraints 95%. This shows at time when fusion AI developing extremely rapidly, it important control within range design appropriate countermeasures. results show nearly 50% choose enterprises, indicating public tends start enterprise side. proportion tend improve requirements reaches 20%, agrees use rational solve possible risks problems governance application issues. Accordingly, completes relevant above dimensions, formulates principles for industry Artificial Intelligence (AI) Dance Art, regulate research

Язык: Английский

Процитировано

0
Attention-Based PSO-LSTM for Emotion Estimation Using EEG DOI Creative Commons

Hayato OKA,

Keiko Ono, Panagiotis Adamidis

и другие.

Sensors, Год журнала: 2024, Номер 24(24), С. 8174 - 8174

Опубликована: Дек. 21, 2024

Recent advances in emotion recognition through Artificial Intelligence (AI) have demonstrated potential applications various fields (e.g., healthcare, advertising, and driving technology), with electroencephalogram (EEG)-based approaches demonstrating superior accuracy compared to facial or vocal methods due their resistance intentional manipulation. This study presents a novel approach enhance EEG-based estimation by emphasizing temporal features efficient parameter space exploration. We propose model combining Long Short-Term Memory (LSTM) an attention mechanism highlight EEG data while optimizing LSTM parameters Particle Swarm Optimization (PSO). The assigned weights hidden states, PSO dynamically optimizes the vital parameters, including units, batch size, dropout rate. Using DEAP SEED datasets, which serve as benchmark datasets for research using EEG, we evaluate model’s performance. For dataset, conduct four-class classification of combinations high low valence arousal states. perform three-class negative, neutral, positive emotions dataset. proposed achieves 0.9409 on surpassing previous state-of-the-art 0.9100 reported Lin et al. attains 0.9732 recording one highest accuracies among related research. These results demonstrate that integrating significantly improves estimation, contributing advancement technology.

Язык: Английский

Процитировано

0