Nearest-Neighbor Gaussian Process
A sparse approximation method for Gaussian processes that restricts the conditional distribution at each location to depend only on its m nearest neighbors, sparsifying the dense covariance matrix and reducing computational complexity from O(n³) to linear.
Timeline (last 90 days)
The variational family of this framework directly mimics the sparse nearest-neighbor autoregressive structure of the prior, allowing the variational approximation to inherit the sparse dependency relationships of the spatial prior.
The Nearest Neighbor Gaussian Process (NNGP) conditions the distribution of each location only on its m nearest neighbors (where m is much smaller than n), thereby sparsifying the dense covariance matrix.
All Facts (2)
The variational family of this framework directly mimics the sparse nearest-neighbor autoregressive structure of the prior, allowing the variational approximation to inherit the sparse dependency relationships of the spatial prior.
50%UnverifiedThe Nearest Neighbor Gaussian Process (NNGP) conditions the distribution of each location only on its m nearest neighbors (where m is much smaller than n), thereby sparsifying the dense covariance matrix.
50%