Research map / Learning theory and algorithms / Submodular and game algorithms
Shoreline: Data-Driven Threshold Estimation of Online Reserves of Cryptocurrency Trading Platforms
Xitong Zhang, He Zhu, Jiayu Zhou
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add temporal random walk to Bi-LSTM embedding | LSTM embedding · IA-Radoslaw-Email | AUC: 0.898 -> 0.905 | Improved |
| add temporal random walk to Bi-LSTM embedding | LSTM embedding · Fb-Forum | AUC: 0.893 -> 0.916 | Improved |
| add temporal random walk to Bi-LSTM embedding | LSTM embedding · Fb-Messages | AUC: 0.787 -> 0.828 | Improved |
| use Bi-LSTM contextual embedding instead of skip-gram node2vec | node2vec · IA-Radoslaw-Email | AUC: 0.840 -> 0.898 | Improved |
| use Bi-LSTM contextual embedding instead of skip-gram node2vec | node2vec · Fb-Messages | AUC: 0.736 -> 0.787 | Improved |
| use deep auto-encoder embeddings (SDNE, DynEmb, GRUEmb) instead of DeepWalk | DeepWalk · IA-Radoslaw-Email | AUC: 0.817 -> 0.761 | Hurt |
| use SDNE instead of DeepWalk | DeepWalk · Fb-Messages | AUC: 0.707 -> 0.769 | Improved |
| add currency embedding features to historical net withdraw LSTM (ablation) | OTE-hist · threshold estimation alpha=10 | loss E: 0.420 -> 0.356 | Improved |
| add raw trading records branch to historical net withdraw LSTM (ablation) | OTE-hist · threshold estimation alpha=10 | loss E: 0.420 -> 0.361 | Improved |
| combine hist, trade and embedding branches (ablation) | OTE-emb · threshold estimation alpha=10 | loss E: 0.356 -> 0.240 | Improved |
| combine hist, trade and embedding branches (ablation) | OTE-emb · threshold estimation alpha=0.01 | loss E: 0.019 -> 0.015 | Improved |
| use learned threshold instead of historical mean | HA · threshold estimation alpha=10 | loss E: 0.753 -> 0.240 | Improved |
Closest papers by meaning
- The Temporal Graph of Bitcoin Transactions (NeurIPS 2025 · similarity 0.83)
- BlockScan: Detecting Anomalies in Blockchain Transactions (NeurIPS 2025 · similarity 0.81)
- EarnHFT: Efficient Hierarchical Reinforcement Learning for High Frequency Trading (AAAI 2024 · similarity 0.81)
- EX-Graph: A Pioneering Dataset Bridging Ethereum and X (ICLR 2024 · similarity 0.81)
- Multi-Chain Graphs of Graphs: A New Approach to Analyzing Blockchain Datasets (NeurIPS 2024 · similarity 0.80)
- MTVHunter: Smart Contracts Vulnerability Detection Based on Multi-Teacher Knowledge Translation (AAAI 2025 · similarity 0.79)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
