Self-Supervised Pretraining
A machine learning method that trains models by designing pretext tasks based on the structure of the data itself, without manual annotation, so that the models learn general representations. It can greatly reduce dependence on labeled data.
Timeline (last 90 days)
Hikvision's AI ultrasonic testing technology achieves generalization across materials and workpieces through self-supervised pre-training and few-shot transfer.
Global classification is not suitable for local tampering detection, and in small-data scenarios, self-supervised pre-training and domain priors are often more effective than larger models.
All Facts (2)
Hikvision's AI ultrasonic testing technology achieves generalization across materials and workpieces through self-supervised pre-training and few-shot transfer.
50%UnverifiedGlobal classification is not suitable for local tampering detection, and in small-data scenarios, self-supervised pre-training and domain priors are often more effective than larger models.
50%