Novelty Detection

Novelty detection aims to identify data points that deviate significantly from a known distribution, a crucial task for robust machine learning in unpredictable environments. Current research emphasizes unsupervised and semi-supervised approaches, employing diverse techniques like contrastive learning, diffusion models, and manifold learning within architectures such as autoencoders and transformers, often incorporating training-free metrics for efficiency. This field is vital for improving the reliability and safety of AI systems across various applications, from autonomous navigation and industrial anomaly detection to cybersecurity and fraud prevention, by enabling systems to adapt to unforeseen circumstances and unexpected data.