id	author	title	date	pages	extension	mime	words	sentences	flesch	summary	cache	txt
work_kjij7pfoonfqzc3h2zbgjvbfpq	Yongqin Xian	Learning from limited labeled data - Zero-Shot and Few-Shot Learning	2020	197	.pdf	application/pdf	80597	8390	72	(DNN) on a large dataset that contains zero-shot test classes also violates the zeroshot learning idea as image feature extraction is a part of the training procedure. We introduce novel (generalized) zero-label and few-label semantic image segmentation tasks in a realistic settings inspired by zero-shot learning for image classification. performance on novel classes in the generalized zero-shot learning setting due our proposed framework is more flexible and can be applied to solve inductive zeroshot learning where there is no image from unseen classes, transductive zero-shot the extreme zero-shot learning case, there is no training images for novel classes. and generalized zero-shot learning setting with various class embeddings (e.g. Mikolov et al., 2013b; Pennington et al., 2014; Miller, 1995) on three challenging unseen class label, i.e. Yts ⊂ Y and in generalized zero-shot learning setting, the In zero-shot learning, class embeddings are as important as image features.	./cache/work_kjij7pfoonfqzc3h2zbgjvbfpq.pdf	./txt/work_kjij7pfoonfqzc3h2zbgjvbfpq.txt