sxhfut / Sentiment-Classification-Via-Galvanic-Skin-Response-Based-on-Deep-Learning-Models

合肥工业大学 计算机与信息学院 情感计算研究所 自然语言处理团队 多模态情感语料库

项目名称

合肥工业大学 计算机与信息学院 情感计算研究所 自然语言处理团队 多模态情感语料库

Sentiment Classification Via Galvanic Skin Response Based on Deep Learning Models

Table of Contents

• Introduction
• MDSTC
• Preprocessed
• Denoising
• Contact

Sentiment Classification Via Galvanic Skin Response

Introduction

We collected a multimodal dataset for sentiment classification, and the dataset is called MDSTC. The Facial Expression of the MDSTC folder contains facial expressions of volunteers watching emotional stimulation clips. The GSR and Pulse File folder contains Galvanic Skin Response and Pulse of volunteers. In this paper, we used Galvanic Skin Response for sentiment classification.

MDSTC

In this section, we show the detail of MDSTC. This image shows the Galvanic Skin Response and Pulse of volunteers watching six types of inspired videos.

This image shows the facial expressions of volunteers watching six types of inspired videos. In the XML file, "Person_info" record the personal information of volunteer, "Film_name" denotes the name of stimulation clips, "PIDIAN" recorded the Galvanic Skin Response, "MAIBO" recorded the pulse.

Preprocessed

If a small amount of data is missing, we fill it with adjacent data. If a large amount of data is missing in a, we delete the data.

Denoising and Data standardization

We use the Butterworth filter to denoise the signal, enhance the useful information, and recover the information degradation caused by the interference. The following image shows the Pulse of a volunteer.

This image shows the raw GSR and GSR after denoising and data standardization of volunteer. Here, you can get more data, link：https://pan.baidu.com/s/1JZCpRxmHl_zeR58AOUrJKg code：pbkr If you need more data, please contact us.

Reference

Xiao Sun, Tao Hong, Changliang Li, Fuji Ren, Hybrid spatiotemporal models for sentiment classification via galvanic skin response, Neurocomputing, Volume 358, 2019, Pages 385-400, ISSN 0925-2312, https://doi.org/10.1016/j.neucom.2019.05.061.

Contact

[孙晓 Xiao Sun] E-mail address: ([e-mail]); Affective Computing and Advanced Intelligent Machine Lab, School of Computer and Information, Hefei University of Technology, Hefei, Anhui, China.
[洪涛 Tao Hong] E-mail address: ([e-mail])