A review of Deep Learning based methods for Affect Analysis using Physiological Signals

Emotions are distinct reactions to internal or external events with implications for the organism. Automatic emotion recognition is a demanding task for pattern recognition and a required information retrieval method for diagnosing the condition of emotions in the peripheral nervous system and psychotherapy. In recent years, scientists have extensively considered physiological signals since they can give a modest, inexpensive, convenient, and easy-to-utilize result for recognizing emotions. Deep Learning has recently demonstrated incredible guarantees in figuring out physiological signals because of its ability to extract useful features and achieve better emotion recognition performance. In this survey, we analyzed a review of the neuro-physiological exploration made from 2012 to 2022, giving a complete outline of the current works in feeling acknowledgment from physiological signals utilizing deep learning strategies. We center our examination on the fundamental viewpoints engaged with the acknowledgment procedure (e.g., stimulus, features extracted, architectures). Our investigation reveals that most researchers have used Convolutional Neural Networks over other deep networks for classifying physiological-based emotions, as deep learning permits automatic end-to-end learning of pre-processing, including extraction and classification components. We determine many challenges and practice suggestions to help the exploration network, especially for the individuals entering this research field.

Automated summary

Emotions are distinct reactions with implications for the organism. Deep Learning has shown great potential in recognizing emotions from physiological signals. Convolutional Neural Networks have been widely used for classifying physiological-based emotions, as they allow automatic end-to-end learning of pre-processing and classification components.