sybil detection 2behavioral similarity 1blockchain analytics 1compression-based methods 1ethereum 1leakage-aware evaluation 1open-world auditing 1sequential models 1transaction modeling 1tree-based models 1
From the 2 of 2 linked papers with an AI index.
2 papers
cs.LG2026
Compression-Based Behavioral Similarity for Open-World Sybil Discovery on Ethereum
MichaÅ Bartnicki, JarosÅaw A. Chudziak
The paper introduces a training‑free, compression‑based similarity approach that builds a symbolic transaction grammar and uses gzip‑based normalized compression distance to identi…
cs.LG2026
Modeling Decisions in Blockchain Analytics: A Leakage-Aware Evaluation of Tree-Based vs. Sequential Models
MichaÅ Bartnicki, JarosÅaw A. Chudziak
The paper proposes a leakage‑aware framework for detecting Sybil bots on Ethereum by comparing tree‑based tabular models with sequence models, showing that XGBoost outperforms Tran…