
CLT for LSS of sample covariance matrices with unbounded dispersions
Under the highdimensional setting that data dimension and sample size t...
read it

Conditional predictive inference for highdimensional stable algorithms
We investigate generically applicable and intuitively appealing predicti...
read it

Risk bounds when learning infinitely many response functions by ordinary linear regression
Consider the problem of learning a large number of response functions si...
read it

Sketching for TwoStage Least Squares Estimation
When there is so much data that they become a computation burden, it is ...
read it

Prediction in latent factor regression: Adaptive PCR and beyond
This work is devoted to the finite sample prediction risk analysis of a ...
read it

A Note on HighDimensional Confidence Regions
Recent advances in statistics introduced versions of the central limit t...
read it

Minimum Description Length Principle in Supervised Learning with Application to Lasso
The minimum description length (MDL) principle in supervised learning is...
read it
Provable More Data Hurt in High Dimensional Least Squares Estimator
This paper investigates the finitesample prediction risk of the highdimensional least squares estimator. We derive the central limit theorem for the prediction risk when both the sample size and the number of features tend to infinity. Furthermore, the finitesample distribution and the confidence interval of the prediction risk are provided. Our theoretical results demonstrate the samplewise nonmonotonicity of the prediction risk and confirm "more data hurt" phenomenon.
READ FULL TEXT
Comments
There are no comments yet.