Emotion Recognition System using IOT and Machine Learning - A Healthcare Application

• Main Authors: Guneet Kaur, Purnendu Shekhar Pandey
• Format: Article
• Language: English
• Published: FRUCT 2018-11-01
• Series: Proceedings of the XXth Conference of Open Innovations Association FRUCT
• Subjects: IoT; Machine learning; Temperature Sensor; Pulse Sensor; ECG Sensor
• Online Access: https://fruct.org/publications/abstract23/files/Kau.pdf
• ISSN: 2305-7254; 2343-0737

Emotion is strong instinctive and intuitive feeling in humans that arise from one’s situations, circumstances, mood and relationship with others. So, emotions are a mental state of person which cannot be stopped. Thus, Emotion Recognition is an important subject when talking about human-machine interaction and when it comes to doing the analysis and controlling of emotions. Emotions can be detected using many techniques like facial expressions, speech, body gesture and psychological signals. Certain physiological changes take place in human body for example, variation in heart rate, temperature, skin conductance, muscle tension and brain waves (ECG). This paper considers temperature and heart beat values (in bpm- beats per minute) extracted from temperature sensor, pulse sensor as well as ECG sensor and use them to train a machine learning algorithm. Naive Bayes Classifier algorithm, a classification problem which is based on conditional probability model and assumes that each feature is independent of the other, the benefit of applying this algorithm lies in the fact that the every individual has different pattern in terms of heart beat and even sometimes the normal body temperature and heartbeat of individuals differ. Also same dataset is used to train K Nearest Neighbors model, another classification problem to help us do comparative analysis of two classification algorithms for instance, which is more accurate, which one is to be preferred when large samples are there, which one is to be preferred when high dimensional is available and so. So, applying Machine Learning will aid in correctly predicting the emotional state of every individual. So, this paper helps to detect emotion of a person and if at all he/ she remains in a sad state or angry state for longer period of time then this product alarms the person to go for consultation. The paper has reached an accuracy of about 84% on sample size of nearly 115 samples.