Tag: few-shot

Researchers developed a lightweight convolutional network combined with modified Reptile meta-learning to accurately recognize sports actions from scarce video samples.

A new meta-learning framework improves concrete strength prediction accuracy and interpretability using limited data.

A decoupled few-shot segmentation model improves edge detail recovery and achieves higher Dice scores across three medical imaging benchmarks.

Improving the efficiency of crop breeding is a critical component of global food security, yet the process remains hindered by ...

A new incremental few-shot fault diagnosis framework uses a diffusion model to generate synthetic samples for rare new fault classes, ...