some questions about the paper and code

[GuChenghs]

Thank you for sharing your paper and code. After I finished reading, I have some questions. Implied in the article,‘For SVM and ridge regression, we observe that keeping meta-training shot higher than meta-testing shot leads to better test accuracies as shown in Figure 2.’ Does this mean that the number of training samples during the meta-train is different from the number of training samples during the meta-test? However，I think in order to better measure the ability of generalization , shouldn't these two parameters be consistent? In addition, when I studied the code, I was confused about ‘train-shot’, ‘val-shot’, ‘train-query’, ‘val-query’, I hope you can explain these. Look forward to your reply.

[kjunelee]

Thanks for your interest in our paper.

Q1: Does this mean that the number of training samples during the meta-train is different from the number of training samples during the meta-test? A1: Yes. The result in Figure 2 is an empirical result and we are not very sure why this phenomenon is happening.

Q2: I was confused about ‘train-shot’, ‘val-shot’, ‘train-query’, ‘val-query’, I hope you can explain these. A2: ‘train-shot’ means number of training examples for each category per episode during meta-training. ‘val-shot’ means number of training examples for each category per episode during meta-validation. ‘train-query’ means number of test (query) examples per episode during meta-training. ‘val-query’ means number of test (query) examples per episode during meta-validation.

[GuChenghs]

Thank you for your reply. When it comes to 'K-way,N-shot',‘ Here K denotes the number of classes, and N denotes the number of training examples per class.’，Implied in the article.Do these mean K-way and N-shot during meta-testing?

[kjunelee]

Yes.