Research talk: Domain-specific pretraining for vertical search
195
Microsoft Research334 тыс
Следующее
25.01.22 – 1732:08Популярные
Опубликовано 25 января 2022, 1:37
Speaker: Tristan Naumann, Principal Researcher, Microsoft Research
Information overload is a prevalent challenge in many high-value domains. Search in biomedicine, and many other vertical domains, is challenging due to the scarcity of direct supervision from click logs. Self-supervised learning has emerged as a promising direction to overcome the annotation bottleneck. In this talk, we discuss work that presents a general approach for vertical search based on domain-specific pretraining and present a case study in the biomedical domain, Microsoft Biomedical Search.
Learn more about the 2021 Microsoft Research Summit: Aka.ms/researchsummit
Information overload is a prevalent challenge in many high-value domains. Search in biomedicine, and many other vertical domains, is challenging due to the scarcity of direct supervision from click logs. Self-supervised learning has emerged as a promising direction to overcome the annotation bottleneck. In this talk, we discuss work that presents a general approach for vertical search based on domain-specific pretraining and present a case study in the biomedical domain, Microsoft Biomedical Search.
Learn more about the 2021 Microsoft Research Summit: Aka.ms/researchsummit
Свежие видео
4 дня – 1 224 44533:33
4 дня – 1 7830:16
7 дней – 7860:50
10 дней – 380 5110:34
10 дней – 1 556 15623:37
10 дней – 9712:14Случайные видео
39 дней – 01:03:41
116 дней – 67 99010:14
157 дней – 4041:00
08.05.21 – 3 4961:39
27.05.20 – 1 2691:04
11.04.19 – 18 9461:44:34
1 день – 820:32
4 дня – 2 1390:44
10 дней – 1 0700:58
14 дней – 103 7672:58
15 дней – 55 0988:15
16 дней – 46839:59
16 дней – 8600:59
11 часов – 3770:55