AOSNET: A neural network implementation of adjustable order statistic filters in fixed time. A neural network, AOSNET, which with an array of binary numbers as input, outputs the kth largest element of the array is proposed. Overall processing time is constant, and does not depend upon the size of the input array, being just seven times the processing time for a single neuron. This neural network may be used as a building block for hardware implementation of order statistic filters. Examples of using AOSNET for implementing the neural networks for sorting binary numbers and for comparison and selection filters (CS filters) are given. A new approach for solving the k-nearest-neighbors problems in Θ(1) time using the sorting neural network principle is described.