A new arXiv paper introduces EthCluster, an unsupervised static analysis method for Ethereum smart contracts. The authors note that poorly designed contracts are especially vulnerable, as attackers can exploit weaknesses to steal the virtual currency those contracts manage.
The method uses unsupervised learning to identify problematic contracts, though the abstract gives few specifics about the underlying technique or dataset. Because the abstract is brief, the paper's full contributions—such as how EthCluster compares with existing tools—remain unclear.
For now, the work signals a growing interest in applying unsupervised learning to smart contract security, potentially offering a way to flag risky contracts without relying on labeled vulnerability examples.