We uncover an ever-overlooked deficiency in the prevailing few-shot learning
(FSL) methods: the pre-trained knowledge is indeed a confounder that limits the
performance. This finding is rooted from our causal assumption: a Structural
Causal Model (SCM) for the causalities among the pre
LSFSL improves the generalizability and robustness of few-shot learning models by incorporating relevant priors and addressing shortcut learning in deep neural networks.