IMPLEMENTASI MODEL DEEP LEARNING INCEPTION RESNET-V2 UNTUK IMAGE CLASSIFICATION PADA MALWARE

Image malware classification is a technique used to identify and classify types of malware using images generated by dynamic and static analysis. The most commonly used method for image malware classification is Convolutional Neural Network (CNN) trained using a dataset of malware images. This research aims to develop a CNN-based image malware classification model using the InceptionResNet-V2 method. The dataset used in this research consists of a total of 9,029 malware image samples from 25 different classes from the Malimg dataset. The testing process is carried out by dividing the dataset into two parts: test data and train data. We use a dataset consisting of various types of malware to train and test our model. We also perform several image augmentation techniques such as flipping and cropping to improve our model's performance. This research runs two different scenarios, scenario 1 uses the original dataset and the other scenario uses a randomly sampled dataset. In this research, the model generated from scenario 1 achieved an accuracy of 87.5% and scenario 2 achieved an accuracy of 85.1% in classifying malware images. This research is expected to contribute to the development of more accurate and effective image malware classification techniques in detecting malware threats on computer systems.