The extensive usage of the Internet and IoT is becoming common in day-to-day life. Advancements of technology and usage of mobile devices in diversified sectors are dominating. Data security is a challenging task when it is shared over Internet. Hackers are constantly posing new threats and introducing new malwares to steal the data, passwords, and vital information. Identification of malwares is essential in this regard. Converting portable executable files into images and applying machine learning and deep learning techniques for the classification of malware type are gaining popularity in the research community. Convolutional neural networks are extensively used in recent times. But they fail in capturing complex patterns in the images. Kolmogorov-Arnold networks are introduced as an alternative to convolutional neural networks as they are able to represent the images in a more discriminable manner. In this work, an algorithm is proposed using Kolmogorov-Arnold network for the classification of malware images. Extensive simulations are done on three malware datasets, viz., MaleVis, Malimg and Microsoft Big 2015. Promising results are obtained with the Kolmogorov-Arnold networks on the three datasets. Proposed algorithm is justified with suitable cross-validation strategies and the results are compared with the state-of-the-art works on similar datasets. Furthermore, a conglomerate dataset is formulated with all the three datasets comprising 60 classes. As all the datasets are imbalanced ones, suitable imbalance metrics, i.e., index balanced accuracy and geometric mean are used. Index balanced accuracies of 98.90, 90.4, 94.06, and 90.1% are obtained on Malimg, MaleVis, Microsoft Big 2015 and Conglomerate datasets, respectively.