Automatic detection of attachment style in married couples through conversation analysis

Koçak, Tuğçe Melike; Dibek, Büşra Çilem; Polat, Esma Nafiye; Kafesçioğlu, Nilüfer; Demiroğlu, Cenk

doi:10.1186/s13636-023-00291-w

EURASIP Journal on Audio, Speech, and Music Processing

Table 6 Accuracy, precision, recall, and F1 scores of the classifiers with low-level acoustic features, i-vector features, and sentiment features are shown using MIM, mRMR, and JMI feature selection algorithms. For each feature selection algorithm, results for only the best performing feature sizes are shown. Feature fusion is used in the second and third columns. Experiments were done using the conversation-based approach. Systems with best F1 scores are shown in bold

From: Automatic detection of attachment style in married couples through conversation analysis

	Low-level acoustic features					Low-level acoustic and I-vector features					Low-level acoustic and sentiment features
	Feature type	Accuracy	Precision	Recall	F1 score	Feature type	Accuracy	Precision	Recall	F1 score	Feature type	Accuracy	Precision	Recall	F1 score
SVM	MIM (50)	0.7	0.69	0.69	0.69	MIM (30)	0.72	0.72	0.68	0.7	MIM (15)	0.68	0.68	0.67	0.68
	mRMR (50)	0.79	0.8	0.76	0.77	mRMR (15)	0.75	0.76	0.72	0.74	mRMR (30)	0.73	0.73	0.74	0.73
	JMI (50)	0.73	0.76	0.65	0.7	JMI (30)	0.7	0.69	0.7	0.7	JMI (30)	0.7	0.7	0.69	0.7
Decision tree	MIM (10)	0.72	0.73	0.69	0.7	MIM (50)	0.74	0.69	0.84	0.75	MIM (5)	0.72	0.76	0.65	0.7
	mRMR (30)	0.73	0.72	0.75	0.74	mRMR (30)	0.73	0.72	0.76	0.74	mRMR (50)	0.73	0.71	0.77	0.74
	JMI (15)	0.72	0.68	0.81	0.74	JMI (5)	0.73	0.72	0.74	0.73	JMI (15)	0.74	0.77	0.69	0.73
Random forest	MIM (30)	0.81	0.82	0.79	0.8	MIM (15)	0.8	0.8	0.79	0.8	MIM (15)	0.8	0.83	0.75	0.79
	mRMR (50)	0.81	0.85	0.75	0.8	mRMR (30)	0.8	0.83	0.76	0.79	mRMR (10)	0.8	0.83	0.75	0.79
	JMI (30)	0.81	0.85	0.75	0.8	JMI (30)	0.82	0.85	0.77	0.81	JMI (15)	0.8	0.83	0.76	0.79
AdaBoost	MIM (30)	0.82	0.81	0.83	0.82	MIM (30)	0.79	0.8	0.77	0.79	MIM (10)	0.78	0.79	0.74	0.77
	mRMR (10)	0.79	0.77	0.81	0.79	mRMR (10)	0.79	0.77	0.83	0.8	mRMR (10)	0.83	0.8	0.86	0.83
	JMI (30)	0.83	0.82	0.84	0.83	JMI (10)	0.8	0.8	0.81	0.8	JMI (30)	0.78	0.78	0.76	0.77
Gradient boosting	MIM (15)	0.8	0.84	0.74	0.79	MIM (30)	0.81	0.82	0.79	0.81	MIM (30)	0.81	0.83	0.77	0.8
	mRMR (15)	0.8	0.81	0.76	0.78	mRMR (30)	0.77	0.8	0.7	0.75	mRMR (30)	0.84	0.88	0.77	0.82
	JMI (15)	0.83	0.84	0.84	0.84	JMI (15)	0.79	0.8	0.76	0.78	JMI (50)	0.8	0.8	0.81	0.8
Extra tree	MIM (15)	0.8	0.8	0.79	0.8	MIM (30)	0.81	0.85	0.76	0.8	MIM (30)	0.82	0.84	0.79	0.81
	mRMR (50)	0.83	0.87	0.79	0.83	mRMR (50)	0.81	0.81	0.81	0.81	mRMR (50)	0.84	0.85	0.81	0.83
	JMI (30)	0.83	0.87	0.8	0.83	JMI (30)	0.84	0.87	0.79	0.83	JMI (15)	0.81	0.82	0.79	0.8
XGBoost	MIM (30)	0.77	0.77	0.76	0.76	MIM (10)	0.75	0.75	0.74	0.75	MIM (10)	0.79	0.77	0.83	0.8
	mRMR (30)	0.8	0.77	0.75	0.76	mRMR (15)	0.75	0.74	0.76	0.75	mRMR (5)	0.78	0.76	0.81	0.78
	JMI (15)	0.8	0.77	0.84	0.8	JMI (10)	0.84	0.83	0.84	0.84	JMI (30)	0.77	0.76	0.77	0.77
Artificial neural network	MIM(50)	0.71	0.72	0.67	0.7	MIM (30)	0.78	0.76	0.79	0.78	MIM(30)	0.76	0.76	0.76	0.76
	mRMR(50)	0.74	0.72	0.76	0.74	mRMR (50)	0.77	0.72	0.86	0.78	mRMR(50)	0.75	0.72	0.81	0.76
	JMI(50)	0.75	0.76	0.72	0.74	JMI (50)	0.71	0.68	0.75	0.72	JMI(50)	0.74	0.78	0.67	0.72
SD-DNN	-	0.8	0.92	0.63	0.75	-	-	-	-	-	-	-	-	-	-

Back to article page