Your browser is out of date

Update your browser to view this website correctly. Update my browser now

×

最新研究:多任务学习

最新的 Cloudera Fast Forward Labs 研究报告和原型、多任务学习,介绍了一种解决机器学习问题的新方法,开启了金融、媒体及医疗保健和农业等各个行业新的可能性。 

与单任务学习不同,多任务学习使得机器学习算法掌握多个任务,从任务之间的关联中受益并产生更准确的模型,更好地推广新任务。而且,由于这些模型可以处理涉及更丰富的现实表现的更复杂挑战,因此它们可以提供跨行业的新科学和商业价值的洞察。 

Newsie是本报告附带的原型,它将新闻文章分类(如,分为新闻、体育、娱乐)。它使用多任务学习进行训练,可正确分类传统正经的大报文章(“任务1”)和更具轰动效应的小报文章(“任务2”)。

Newsie的文章视图显示了模型如何逐字逐级地进行分类。从上下文中的句子到出版地,语境改变了词义。通过多任务学习,Newsie为用户提供了一个窗口,可以了解跨传统严肃和追求轰动的媒体在覆盖和语言使用方面的差异;将当前的新闻转化为视角。

Fast Forward Labs 研究入门

Yes, I would like to be contacted by Cloudera for newsletters, promotions, events and marketing activities. Please read our privacy and data policy.
Yes, I consent to my information being shared with Cloudera's solution partners to offer related products and services. Please read our privacy and data policy.

I agree to Cloudera's terms and conditions.

Your form submission has failed.

This may have been caused by one of the following:

  • Your request timed out
  • A plugin/browser extention blocked the submission. If you have an ad blocking plugin please disable it and close this message to reload the page.