Abstract：
Traditional media such as text, images, audio, and video primarily target specific senses like vision and hearing. In contrast, multiple sensorial media aims to create immersive experiences by integrating additional sensory modalities such as touch, smell, and taste where applicable. Tactile enhanced audio-visual content leverages the sense of touch in addition to visual and auditory stimuli, aiming to create a more immersive and engaging interaction for users. Previously, tactile enhanced content has been explored in 2D emotional space (valence and arousal). In this paper, EEG data against tactile enhanced audio-visual content is labeled based on a self-assessment manikin scale in 3 dimensions i.e., valence, arousal, and dominance. Statistical significance (with a 95% confidence interval) is also established based on gathered scores, highlighting a significant difference in the arousal and dominance dimension of traditional media and tactile enhanced media. A new methodology is proposed using classifier-dependent feature selection approach to classify valence, arousal, and dominance states using three different classifiers. A highest accuracy of 75%, 73.8%, and 75% is achieved for classifying valence, arousal, and dominance states, respectively. The proposed scheme outperforms previous emotion recognition based studies in response to enhanced multimedia content in terms of accuracy, F-score, and other error parameters.
摘要：
文字等传统媒体,images,音频,视频主要针对特定的感官，如视觉和听觉。相比之下,多种感官媒体旨在通过整合其他感官方式（如触摸）来创造身临其境的体验，气味,并在适用的地方品尝。触觉增强的视听内容除了视觉和听觉刺激外，还利用触觉，旨在为用户创建更具沉浸感和吸引力的交互。以前,触觉增强内容已在2D情感空间(效价和唤醒)中进行了探索。在本文中,针对触觉增强视听内容的EEG数据基于3维的自我评估人体模型量表进行标记，即价，唤醒,和优势。统计显著性(95%置信区间)也是基于收集的分数来建立的，强调了传统媒体和触觉增强媒体在唤醒和优势维度上的显著差异。提出了一种使用依赖于分类器的特征选择方法对效价进行分类的新方法，唤醒,和使用三个不同分类器的优势状态。最高准确率为75%，73.8%,75％用于对化合价进行分类，唤醒,和优势国家，分别。所提出的方案在准确性方面优于先前基于情感识别的研究，以响应增强的多媒体内容，F分数,和其他错误参数。
