学术报告：Transfer Learning from Possibly Dissimilar Sources: A Source-Function Weighted Method

报告时间：6月26日（星期三）上午10:00-11:30

报告地点：沙河校区，二教109

报告人：林路，山东大学金融研究院，教授

报告摘要：The relatedness or similarity between source domains and a target domain seems to be essential to a positive transfer learning. In practice, however, the similarity condition is difficult to check and is often violated. In this paper, instead of the popularly used similarity condition, a seeming similarity is introduced, which is defined by a non-orthogonality together with a smoothness. Such a condition is naturally satisfied under common situations and even contains the dissimilarity as its special case. Based on the seeming similarity and an L2-adjustment, a source-function weighted-transfer learning estimation (sw-TLE) is constructed. By source-function weighting, the transfer learning adaptability is achieved in the sense that the transfer learning strategy is always positive in both similar and dissimilar scenarios. Particularly, under the case with homogenous sources, the sw-TLE even obtains the parametric or semiparametric convergence rate, though the model under study is actually nonparametric. The hidden relationship between the source-function weighting estimator and the James-Stein estimator is established as well, which reveals the structural reasonability of our methodology. Moreover, the strategy does apply to nonparametric and semiparametric models. The comprehensive simulation studies and real data analysis can illustrate that the new strategy is significantly better than the competitors, and is comparable with the oracle estimator.

报告人简介：林路是山东大学中泰证券金融研究院教授、博士生导师，第一和第二届教育部应用统计专业硕士教育指导委员会成员，山东省教育厅应用统计专业硕士教育指导委员会成员，山东省政府参事，济南应用数学高等研究院院长。从事大数据、高维统计、非参数和半参数统计以及金融统计等方的研究，在国内外统计学、机器学习和相关应用学科顶级期刊和重要期刊(包括Ann. Statist., JMLR, Stat. Comput.和中国科学)发表研究论文130余篇；多个金融策略资政报告得到省长的正面批示；主持过多项国家自然科学基金课题、全国统计科学研究重大项目、教育部博士点专项基金课题、教育部新文科课题、山东省自然科学基金重点项目等；获得国家统计局颁发的全国统计优秀研究成果一等和二等奖，山东省优秀教学成果一等奖(均排名第一)。