Ivaylo Penev, Milena Karova, Mariana Todorova
machine learning, nearest neighbors, kNN, hand-written digits, recognition, classification
The paper presents application of the k-nearest neighbor algorithm (kNN) for recognition of hand-written digits from 0 to 9. The emphasis is on the choice of the number of the nearest neighbors (the k parameter), which has significant impact on the algorithm performance. The main steps of the algorithm are described. The function for distance calculation and the method for choosing a class of the recognized digit are explained. Experimental results are presented. According to the results recommendations for the choice of k are summarized. The aim is increasing the performance of the kNN algorithm for the hand-written digit recognition problem, regarding two criteria – percent of the correctly recognized input data and time for recognition.
Cite this paper
Ivaylo Penev, Milena Karova, Mariana Todorova. (2016) Exploration of the K Parameter in Hand-Written Digit Recognition by K-Nearest Neighbor Algorithm. International Journal of Control Systems and Robotics, 1, 158-161