This paper investigates the effects of collaborative filtering on the classification of corrupted digit images. Several experiments were carried out using the MNIST digit dataset by applying a corruption model to the original images, two reconstruction algorithms, and an SVM classifier to measure classification accuracy at each stage of processing. The results demonstrate that collaborative fil...