Sparse Covariance Matrix Estimation With Eigenvalue Constraints.

Area of Science:

• Statistics
• Machine Learning
• Data Science

Background:

• Estimating covariance matrices is crucial in high-dimensional data analysis.
• Existing methods may struggle to guarantee positive definiteness and sparsity simultaneously.
• High-dimensional covariance estimation presents significant statistical and computational challenges.

Purpose of the Study:

• To develop a new method for estimating high-dimensional, positive-definite covariance matrices.
• To ensure the estimated matrix possesses both sparsity and positive definiteness.
• To provide a statistically and computationally efficient estimation procedure.

Main Methods:

• Extension of the generalized thresholding operator with an explicit eigenvalue constraint.
• Development of an iterative soft-thresholding and projection algorithm.
• Algorithm based on the alternating direction method of multipliers (ADMM).

Main Results:

• The proposed estimator achieves simultaneous sparsity and positive definiteness.
• The estimator is proven to be rate optimal in the minimax sense.
• Numerical experiments demonstrate the method's effectiveness on simulated and real data.

Conclusions:

• The new approach effectively estimates high-dimensional covariance matrices.
• The method guarantees desirable properties like sparsity and positive definiteness.
• The developed algorithm is efficient and practical for real-world applications.