主  題：Spline LASSO in high-dimensional linear regression

內(nèi)容簡介：A Spline-LASSO approach is proposed for a high-dimensional linear regression problem, where the covariates are ordered in a meaningful way. It is designed to tackle the case where the shape of the parameter values changes smoothly, whereas fused Lasso is better fitted for piecewise constant functions. Computationally it can be easily modified to use LARS algorithms. Simulations and real example will be given.

報告人：荊炳義    教授    博導(dǎo)    教育部長江講座教授

時  間：2016-11-04    14:30

地  點：競慧東樓305

舉辦單位:理學(xué)院  統(tǒng)計與大數(shù)據(jù)科學(xué)研究院  科研部