A New Framework for Fraud Detection in Bitcoin Transactions Through Ensemble Stacking Model in Smart Cities
Bibliographic record
Abstract
Bitcoin has a reputation of being used for unlawful activities, such as money laundering, dark web transactions, and payments for ransomware in the context of smart cities. Blockchain technology prevents illegal transactions, but cannot detect these transactions. Anomaly detection is a fundamental technique for recognizing potential fraud. The heuristic and signature-based approaches were the foundation of earlier detection techniques, but tragically, these methods were insufficient to explore the entire complexity of anomaly detection. Machine Learning (ML) is a promising approach to anomaly detection, as it can be trained on large datasets of known malware samples to identify patterns and features of the transactions. Researchers are focusing on determining an efficient fraud and security threat detection model that overcomes the drawbacks of the existing methods. Therefore, ensemble learning can be applied to anomaly detection in Bitcoin by combining multiple ML classifiers. In the proposed model, the ADASYN-TL (Adaptive Synthetic + Tomek Link) balancing technique is used for data balancing. Random search, grid search and Bayesian optimization are used for hyperparameter tuning. The hyperparameters have a great impact on the performance of the model. For classification, we used the stacking model by combining Decision Tree, Naive Bayes, K-Nearest Neighbors, and Random Forest. We used SHapley Additive exPlanation (SHAP) to interpret the predictions of the stacking model. The model also explores the performance of different classifiers using accuracy, F1-score, Area Under Curve-Receiver Operating Characteristic (AUC-ROC), precision, recall, False Positive Rate (FPR) and execution time, and ultimately selects the ideal model. The proposed model contributes to the development of effective fraud detection models that address the limitations of the existing algorithms. Our stacking model, which combines the prediction of multiple classifiers, achieved the highest F1-score of 97%, precision of 96%, recall of 98%, accuracy of 97%, AUC-ROC of 99% and FPR of 3%.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.002 | 0.003 |
| Meta-epidemiology (narrow) | 0.001 | 0.001 |
| Meta-epidemiology (broad) | 0.002 | 0.002 |
| Bibliometrics | 0.002 | 0.001 |
| Science and technology studies | 0.001 | 0.001 |
| Scholarly communication | 0.002 | 0.003 |
| Open science | 0.003 | 0.001 |
| Research integrity | 0.002 | 0.002 |
| Insufficient payload (model declined to judge) | 0.003 | 0.000 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".