Examine Manipulated Datasets with Topology Data Analysis: A Case Study

摘要: Learning and mining technologies have been broadly applied to reveal the value of tremendous data impact decision-making. Usually, correctness decisions roots in truth for these technologies. Data fraud presents everywhere, even if were true, could be maliciously manipulated by cyber-attackers. Methods long exploited examine authenticity, but are less effective when only values without violating scopes definitions. Then made from wrong or hijacked. It has concluded that manipulation is latest technique “the art war cyberspace.” Examining each instance its source exhaustive impossible, example recollecting national consensus. In this paper, through a case study on banknotes, we exploit Topological Analysis (TDA) examining data. A fraction records examined integrally other than individually. The possibility using TDA verify efficiently then evaluated. We first test above detection, discuss limitations state art. Although not so matured, it reported many applications, now our work evidences usage anomalies.