1: \begin{abstract}

2:     We propose estimating \emph{Gaussian graphical models (GGMs)} that are \emph{fair with respect to sensitive nodal attributes}. 

3:     Many real-world models exhibit unfair discriminatory behavior due to biases in data.

4:     Such discrimination is known to be exacerbated when data is equipped with pairwise relationships encoded in a graph.

5:     Additionally, the effect of biased data on graphical models is largely underexplored. 

6:     We thus introduce fairness for graphical models in the form of two bias metrics to promote balance in statistical similarities across nodal groups with different sensitive attributes. 

7:     Leveraging these metrics, we present Fair GLASSO, a regularized graphical lasso approach to obtain sparse Gaussian precision matrices with unbiased statistical dependencies across groups.

8:     We also propose an efficient proximal gradient algorithm to obtain the estimates.

9:     Theoretically, we express the tradeoff between fair and accurate estimated precision matrices.

10:     Critically, this includes demonstrating when accuracy can be preserved in the presence of a fairness regularizer.

11:     On top of this, we study the complexity of Fair GLASSO and demonstrate that our algorithm enjoys a fast convergence rate.

12:     Our empirical validation includes synthetic and real-world simulations that illustrate the value and effectiveness of our proposed optimization problem and iterative algorithm.

13: \end{abstract}

14: