Abstract: ‍ ‍Due to the decreasing recognition rate of the existing small sample recognition methods ignoring the different effects of various task experiences on new tasks， a meta-learning recognition method with small sample was proposed for high resolution range profile （HRRP） of ship targets， based on loss weighted correction. Based on the meta-learning theory， a pre-training model composed of a basic learner and a meta learner was designed. The characteristic loss of different tasks exhibits difference degrees of learning experience， hence the loss function of meta learner was weighted by the loss， to reduce the biased impact from different tasks. By using the learning experience of simulation data， resulted from the pre-training model， the recognition of measured HRRP was carried out for ship targets， on a small sample test task set. Experimental results show that， the proposed method outperforms the existing contrastive models， especially in the case of smaller samples， in the view of classification and recognition abilities.