Precision Transfer Learning via Correlation Ratio Combination between Target and Heterogeneous Sources

发布时间：2021-11-08 作者：浏览次数：

Speaker:	林路	DateTime:	2021年11月9日（周二）上午10：00 - 11：00
Brief Introduction to Speaker: 林路，山东大学。		Place:	腾讯会议，会议号693 639 223
Abstract:A basic condition for efficient transfer learning is the similarity between the target model and source models.In practice, however, the similarity condition is difficult to meet or is often violated. Instead of the similarity condition, a bran-new strategy, linear correlation ratio, is introduced in this paper under the framework of exponential family with heterogeneous sources. Such a correlation ratio can accurately combine target with sources, and can be easily estimated by historical data or historical characteristics of permanent variables.A precision transfer learning likelihood is then constructed via the correlation ratio. On the practical side, the new framework is applied to some application scenarios, specially the area of data streams.