Mouse Dynamics_to_Image

Research on mouse dynamics has largely suffered from a higher accuracy threshold relative to keystroke dynamics. Exploring the potential of deep learning algorithms as an alternative has also largely suffered from limited datasets and applicability. To address this underlying limitation, this study developed a novel dataset that captured the temporal and spatial components of the existing mouse dynamics dataset (Balabit: https:\/\/github.com\/balabit\/Mouse-Dynamics-Challenge) to create an image dataset. Extensive image and signal processing techniques can then be leveraged to improve the accuracy of mouse dynamics, both within the authentication and attribution studies of cybersecurity. Diverse timeseries to image conversion techniques, which include GAF, MTF, Spectrogram, Heatmap, and WaveletTransform, were used to curate this dataset.