This project employs CNN to classify malware based on their images.

Using the malimg dataset of grayscale malware images, the goal of this project is to develop an accurate model for categorizing different types of malware. By leveraging shared visual characteristics, the approach represents malware binaries as grayscale images and groups similar types of malware based on layout and texture similarities within families.

• Malware image classification using Convolutional Neural Networks (CNN)
• Dataset preprocessing and augmentation techniques
• Baseline Logistic Regression (LR) model for comparison
• Evaluation metrics for performance analysis
• Heatmap visualization for interpretability and explainability

• Python 3.x
• Required Python packages (specified in requirements.txt)

1. Clone the repository: git clone https://github.com/vaishnaviu/Malware-Image-Classifier-with-CNN.git cd Malware-Image-Classifier-with-CNN

2. Install the required Python packages: pip install -r requirements.txt

We used an altered version of the malimg dataset, which includes labeled malware samples. It has been augmented with non-malware images to create a ”normal” class. This enabled comprehensive evaluation of the models’ performance. The malimg dataset contains grayscale images of malware binaries from 25 different families. Each image is a 32x32 byteplot representation of the malware. The altered malimg dataset and additional dataset used for testing can be obtained from the following sources:

• Malimg Dataset
• Additional Dataset

1. Prepare the dataset by following the instructions provided in the repository.
2. Run the main.py file to train and evaluate the models: python main.py

The trained CNN model achieved an accuracy of 98% on the testing set, outperforming the baseline Logistic Regression model. Heatmaps were generated to visualize important regions in the images that contribute to the model's predictions.

• Malimg Dataset
• Research papers and references:

[1] ”Malware Classification Using Convolutional Neural Networks,” by S. S. Gandhi et al.

[2] Malware Images: Visualization and Automatic Classification from L. Nataraj, S.Karthikeyan, G. Jacob and B. S. Manjunath

[3] A Review on Machine Learning Approaches for Malware Detection and Classification (Bajaj et al., 2019).

[4] Symantec Global Internet Security Threat Report, April 2010.

[5] Malware Detection Based on Convolutional Neural Networks (Li et al., 2018).