CARLA: Self-supervised contrastive representation learning for time series anomaly detection

Zahra Zamanzadeh Darban, Geoffrey I. Webb, Shirui Pan, Charu C. Aggarwal, Mahsa Salehi

Abstract: Highlights•CARLA is a novel contrastive learning-based time series anomaly detection framework.•CARLA uses anomaly injection to create negative samples for contrastive learning.•CARLA’s effectiveness is validated through extensive and robust evaluation approach.