search

csyanbin / TPN

Code for ICLR19 paper: Learning to Propagate Labels: Transductive Propagation Network for Few-shot Learning.
243 stars 43 forks source link

Question about query sample numbers #5

Closed ZhenLiuBuaa closed 4 years ago

ZhenLiuBuaa commented 4 years ago

Hello, it's really a nice work. And i found when 5*1 test samples, the model has a good performance. But i wonder if there is only one query sample, does the model has a recognize ability and how to test it?

csyanbin commented 4 years ago

Our assumption is that we can make use of the unlabelled query examples to gain better performance. The one query you mentioned is the so-called inductive in our paper.

We didn't test this setting. If you want to make some experiments, you can simply change the query examples in dataloader or model.py. This should be very simple.