Screening Emotional Physiological Signal Features Using Ant Colony Optimization

Emotions play a significant role in human perception and decision making. Along with the development of artificial intelligence these years,the combination of emotion intelligence and computer technology brings the novel research area named affective computing.This combination will greatly advance the development of the computer technology.With the people continuously explore the cognitive field,the research of emotional state field will be future development.At present,the signals of emotion recognition mostly using facial expressions,body posture and sound signals, the results of these methods are often influenced by subjects,which cannot completely reflect the real emotional state.The physiological signals directly controlled by the human autonomous nervous system and incretion system,not by subjects.Therefore, using physiological signals as the research objects of emotion recognition,the data will be more objective,also could better recognize the internal emotion and affective changes.Picard and colleagues at MIT Media Laboratory firstly extracted the features from physiological signals in researching emotion recognition,had testified that it is feasible.However,recent studies on emotion recognition based on physiological signals still have some problems.Such as,cannot be get the effective feature combination,the recognition rate is not high enough to be widely used in practice, robustness is poor,and so on.Therefore,research on physiological signal's emotion recognition still has larger developing spatial.Especially,the feature select of signals is an important problem in emotion recognition.Ant Colony Optimization(ACO) algorithm is new simulated evolutionary algorithm,through simulate the action of nature ants paths searched in foraging.The algorithm not only has a good robustness and positive feedback characteristics,but also has the features of juxtaposition distributed computing.Moreover,Ant Colony Optimization has better execution efficiency and convergence speed than other optimization algorithm on condition of never prior knowledge.This paper tries to use ACO algorithm to select features from emotion physiological signals.The main contents and achievements are listed as follows:(1) Physiological signals data acquisition:experiments collect Conner's ECG signals in joy and sadness by physiological signal recorder(MP150),and analyze the random 150 group features.(2) Physiological signals feature extraction:make original ECG signals discrete wavelet transform,wipe off out of band noises and baseline drift,and obtain the high frequency part of ECG signals,compute statistical features;extract the representative features of ECG signals,form the original feature matrix.(3) Physiological signals feature select and classification:use ACO algorithm and k-Nearest Neighbor classification algorithm to select features,recognize joy and sadness emotional state,then obtain the most effective feature combination of classified emotional state.(4) Simulate the emotion database of Augsburg University,for testing the feasibility of physiological signals emotion recognition by using ACO algorithm and kNN classification.Analyze the parameter setting of Ant Colony Optimization,compared the result with other intelligent optimization algorithm,to obtain the most effective feature subsets in different emotion recognition.