通過生理分析使用者情緒

Abstract

現今大多數的穿戴式智慧裝置著重於人體行為辨識，例如慢跑、騎腳踏車、游泳，甚至是睡眠。這些發明幫助我們檢視自己的身體狀況也使我們保持健康。但實際上影響我們健康的原因，不僅僅只是形於外的肢體活動，人們每天的精神狀態及情緒起伏其實也包含在其中。這幾種因素環環相扣，顯著地影響人類行為與身體健康。於是這幾年來，有越來越多研究者開始投入心力在情感認知上。於此篇論文中，我們使用現成的穿戴式感應器，例如心跳率、皮膚導電反應和體感溫度偵測來解讀使用者的生理訊號，並運用機器學習技術來分析受測者的情緒狀態。這些情緒狀態是能提升身理健康和human-machine interaction的關鍵因素。我們對實際情境進行了實驗並假設共有三種情緒狀態，由數據結果得知，此試驗達到相當高的識別準確率，有97.31％ 。除此之外，我們開發一個Android系統的應用程式，它可以透過藍芽裝置與實驗系統連接，便可即時了解受測者的情緒狀態。

Most of the existing smart bands focus on physical activities recognition, such as running, cycling and even sleeping. They help us to monitor our physical health and keep ourselves healthy. But what influences our health is not only physical activities, it also includes mental and emotional states that we experience throughout the day. These states build our behavior and affect our physical health significantly. Therefore, emotion recognition draws more and more attention of researchers in recent years. In this thesis, we propose a system that uses off-the-shelf wearable sensors, such as heart-rate, galvanic skin response, and body temperature sensors, to read physiological signals from the user and apply machine learning techniques to recognize emotional states of the user. These states are key steps towards improving not only physical health but also emotional intelligence in advanced human-machine interaction. We consider three types of emotional states and conducted experiments on real-life scenarios, achieving highest recognition accuracy of 97.31%. Furthermore we have developed Android application that connects to our prototype via Bluetooth and detects users’ emotional state in real-time fashion.