Efficiently measuring similarity between pieces of arbitrary binary data is a useful capability in many applications. Similarity digest schemes are a family of tools for this task. They first map all inputs to usually small sized similarity digests, then make comparisons based on these digests alone. While these schemes proved to be very effective in some fields like malware analysis and digital forensics, the study on their robustness against targeted attacks, which is crucial in said applications, remains limited. The student's task is to enumerate existing attacks described in literature, and implement them in a common framework that allows for a comperative understaning over the adversarial robustness of existing schemes and their underlying methods.