Enhancing Cryptocurrency Fraud Detection with Hybrid Graph-Temporal Neural Networks

Decentralisation is the next booming thing. One of the major applications of decentralisation is cryptocurrencies which are deployed on a blockchain architecture. Almost ev-eryone in the world has been introduced to cryptocurrency due to its massive outreach. It is a form of currency but in digital form and way more valuable. Cryptocurrency has amassed a lot of young followers due to its high lucrative returns. With Bitcoin reaching new peaks, it has directly put cryptocurrencies in cross hairs of hustlers who want to fraud their way into wealth. The rate of fraud cases in crypto markets has been increasing linearly. To avoid such cases this paper will propose a novel deep neural network architecture called CryptoFraudNet which will make use Graphs components, Self Attention mechanism, etc. to capture even the smallest discrepancies in data.