API
< Back to all clusters
[TECHNOLOGY] · South Korea, Italy · 2 sources

started · updated

Dunamu presents AI blockchain research at ICAIF 2026

Dunamu, the operator of the Upbit cryptocurrency exchange, has had three research papers accepted at the ACM International Conference on AI in Finance (ICAIF 2026) held in Milan, Italy. The research focuses on utilizing artificial intelligence to enhance the analysis of blockchain data.

One study introduces a machine learning methodology to identify exchange-owned wallets by analyzing transaction patterns, token movements, and gas usage. By analyzing approximately 3.03 million Ethereum addresses using Proof-of-Reserves data from 15 exchanges, the model achieved high accuracy in identifying wallets, even those not included in the initial training set. This technology is already being integrated into Upbit Data Lab services.

Two additional workshop papers address digital asset classification and fund tracking. One method employs Large Language Models (LLMs) to interpret smart contract code into natural language, allowing for the classification of decentralized finance (DeFi) protocols based on their actual financial functions. The other study examines the reliability of estimating fund usage when blockchain transaction records are incomplete. Dunamu plans to apply these AI-driven methodologies to refine its digital asset classification systems and on-chain fund tracking capabilities.

Entities

ACM · Dunamu · ICAIF 2026 · Milan · Upbit