Tuesday, November 17, 2026

Three Dunamu Inc. Papers Accepted at International Financial AI Conference

Input
2026-11-17 13:50:26
Updated
2026-11-17 13:50:26
Dunamu Inc.'s corporate identity. Provided by Dunamu Inc.

[Financial News] Three research papers on artificial intelligence (AI) and blockchain by Dunamu Inc., which operates the virtual asset exchange Upbit, have been accepted for the international financial AI conference “ICAIF 2026.” The technology for identifying exchange wallets among them has been put to practical use in Upbit Data Lab.
According to the virtual asset industry on the 17th, Dunamu Inc. had one main conference paper and two workshop papers accepted at ICAIF 2026, held in Milan, Italy. More than 1,000 papers were submitted to the conference, and the acceptance rate for the main conference was about 22%.
The main conference paper examined AI-based technology for identifying exchange wallets. The researchers analyzed about 3.03 million Ethereum wallet addresses using proof-of-reserves data from 15 exchanges. In a performance evaluation of exchange wallets not used for training, the system recorded a PR-AUC score of 0.9594.
The other two papers addressed virtual asset classification and on-chain fund tracing. In a classification experiment using a large language model (LLM), 16 of the 17 items analyzed matched existing criteria. Another paper proposed conditions for analyzing fund flows from incomplete transaction records.
Damon Kim, Dunamu Inc.’s chief data officer (CDO), said the research demonstrated the potential to improve the reliability of AI-based blockchain data analysis and that research into related technologies would continue.
Meanwhile, Dunamu Inc. is incorporating AI into its virtual asset analysis services, including by adding the AI-based market information services “AI Market Outlook Report” and “Uptrace” to Upbit Data Lab last month.

[email protected] Im Sang-hyeok Reporter