Vinsamlegast notið þetta auðkenni þegar þið vitnið til verksins eða tengið í það: http://hdl.handle.net/1946/39359
Money laundering is a serious problem in the world today and especially in the new and growing cryptocurrency market. This is mostly because of the anonymity that many cryptocurrencies offer and the fact that transactions between two actors don't require a third party to make the transaction. That is why cryptocurrencies are a haven for money laundering, because it is easier for criminal entities to buy the currency on the deep web or other places and then trade the crypto for real fiat money. To hide their paths they can send the coins into mixers provided by mixing services that create complicated transaction paths that make it almost impossible for humans to track. Therefore detecting money laundering in cryptocurrency can be tricky, not just because of the anonymity in crypto, since you often don't know the identity of actors in transactions, but because the crypto network is large and convoluted and nearly impossible to analyze by hand. What we on the other hand can do is look at addresses that took part in transactions as actors and then use machine learning to predict what addresses are possibly laundering money or exhibiting some deviations of the norm. \\
In this paper we intend to analyse methods we believe can be used to detect money laundering in cryptocurrencies using machine learning to empower investigators to more accurately and efficiently determine whether suspicious activity is money laundering.