0


在线广告拍卖中自利数据源的优化学习

Learning optimally from self-interested data sources in on-line ad auctions
课程网址: http://videolectures.net/bsciw08_zoeter_lofsdsioao/  
主讲教师: Onno Zoeter
开课单位: 施乐欧洲研究中心
开课时间: 2019-10-31
课程语种: 英语
中文简介:
在对当前在线广告拍卖的分析中,人们通常假定点击率等基本参数是已知的。对现实中存在的不确定性的忽视导致了几个严重的问题。在今天的演讲中,我们将着重强调两点:(1)对新广告没有原则性的探索;(2)广告商没有动力只订阅目标明确的关键词。事实上,这是一个非常有趣的机会,目标非常低的广告商利用这一事实。我们提出了一个新的拍卖,解决了这两个问题。这次拍卖的关键技巧是,广告商不仅要提交出价,还要相信自己的点击率。
课程简介: In the analysis of current online ad auctions essential parameters such as click-through rates are often assumed to be known. The disregard of the uncertainty that is present in reality leads to several serious problems. In the talk we will highlight two: (i) there is no principled exploration of new ads, and (ii) there is no incentive for advertisers to only subscribe to well targeted key-words. In fact, there is an interesting opportunity for very poorly targeting advertisers to exploit this fact. We present a new auction that solves both problems. The key trick for this auction is that advertisers are not only requested to submit a bid, but also a belief over their own click-through rate.
关 键 词: 在线广告; 拍卖; 自利数据源的优化; 学习
课程来源: 视频讲座网
最后编审: 2019-10-29:lxf
阅读次数: 25