Sponsored by Zettablock
A vast majority of exploits rely on malicious smart contracts. The existing malicious smart contract detection bot does a great job of identifying malicious smart contracts. It relies on a supervised machine learning model (logistic regression) trained on the byte code of thousands of smart contracts. Can it be improved to further increase precision/ recall?
Build a new machine learning model that outperforms the existing model (10% increase of F1-score). Note, the current model has been trained on Ethereum contracts and should be extended to BSC and Polygon contracts. The emitted alerts should match those of the existing malicious smart contract detection bot. Given there are only few known malicious smart contracts, first the smart contract structure should be learnt with an unsupervised model; a supervised model with labels should be trained in a second step.
The deliverables of this bounty should be:
In return for developing a functional Forta detection bot that alerts on the above instance, the Forta Foundation is offering $5,000 in FORT, the native utility token of the Forta Network. The Forta Foundation team will review the final results and assign the rewards accordingly. The Forta Foundation will cover initial Bot deployment costs (including staking).
In order to ensure that all the work funded by the Forta Foundation under the Threat Research Initiative is made available for the benefit of Forta users, ownership over any bot that receives rewards will be assigned to the Foundation, which will be made available to the public under the "Forta Bot License”. Participants should also commit to making suggested improvements during the first 2 months after deployment.
You are a data scientist who is familiar with or experienced in bot development in the Forta Network. You can communicate effectively with the team, provide updates, and support precision and recall testing of the bot.
*Forta is a real-time detection network for security & operational monitoring of blockchain activity secured by FORT token. Billions of dollars in Web3 monitored by Forta. Underlying Forta is a decentralized network of independent node operators that scan all transactions and block-by-block state changes for outlier transactions and threats. When an issue is detected, node operators send alerts to subscribers of potential risks, which enables them to take action.*
*Forta Detection Bots are a set of code scripts within a Docker container that process some blockchain data (i.e. a block or transaction) and detect specific threat conditions (e.g. whether a flash loan attack occurred, or whether a particular account balance fell below some threshold). Bots emit alerts for their findings. Bots are executed by scan nodes.*