TDA – comparing and contrasting the persistence diagrams of different software

The application of TDA and its technique of persistent homology offers a unique approach to differentiating benign software from malicious ones (malware), even amid the complexity and noise present in high-dimensional datasets.

Let’s delve into this by further expanding on the examples provided. First, consider benign software – programs designed to perform legitimate, useful tasks without causing harm to the system. When subjected to TDA, the properties of benign software tend to form certain predictable patterns. These properties, which can include binary structures, system calls, or network activity, may cluster together in the topological space. This is like how people at a social gathering might group based on shared interests or common connections. In terms of our earlier analogy, these clusters can be viewed as “mountains” on our landscape.

In the context of...