id	author	title	date	pages	extension	mime	words	sentences	flesch	summary	cache	txt
work_mxyrjrwatzbxze2i7rhlholzum	Yuan Zhang	Aspect-augmented Adversarial Networks for Domain Adaptation	2017	14	.pdf	application/pdf	8589	908	62	learning between two (source and target) classification tasks or aspects over the same domain. sentence-level aspect relevance to learn how to encode the examples (e.g., pathology reports) from the can be adjusted only based on the source class labels, and that it also reasonably applies to the target encodings, we must align the two sets of encoded examples.2 Learning this alignment is posinvariant representation, we introduce an adversarial domain classifier analogous to the recent successful use of adversarial training in computer vision (Ganin and Lempitsky, 2014). in our approach, 1) aspect-driven encoding, 2) classification of source labels, and 3) domain adversary, (aspect transfer) as well as on a more standard review dataset (domain adaptation). Domain Adaptation for Deep Learning Existing approaches commonly induce abstract representations without pulling apart different aspects in the methods first learn a task-independent representation, and then train a label predictor (e.g. SVM)	./cache/work_mxyrjrwatzbxze2i7rhlholzum.pdf	./txt/work_mxyrjrwatzbxze2i7rhlholzum.txt