Lakshmi P. Krishnan, Iman Vakilinia, Sandeep Reddivari, Sanjay Ahuja
Blockchain technology and cryptocurrencies have captured global attention due to their numerous and versatile features, resulting in several industries and services adopting cryptocurrencies as payment methods. The advantages including user anonymity, open source, and tamper-proof transactions, have contributed to its popularity. However, these advantages have also attracted scammers who exploit the technology's features to engage in fraudulent activities, leading to a growth in crypto-frauds. To prevent and identify these frauds, various detection and prevention methodologies have been proposed, mainly using machine learning algorithms to identify scams as anomalies or outliers. The performance of such models depends on the datasets used and the features engineered. Often, these models face the challenge of having limited amounts of data that are scam-labeled. Under such circumstances, the model performs poorly due to an imbalanced dataset. Similarly, the features engineered and selected to train the model are also very important in detecting scams. With the help of sampling techniques, we propose to create a dataset that is researchready and addresses the data imbalance problem. Additionally, we list the resources that can be used to collect labeled data. Furthermore, we discuss the practical significance of features and various feature engineering strategies in detecting scams from transactions in Ethereum.
In recent times, blockchain technology, especially Ethereum, has gained tremendous popularity for supporting various financial transactions. Ethereum, the second most prominent cryptocurrency platform after Bitcoin, processes over one million transactions daily, known for its high security and ease of use. However, despite its robust security features, Ethereum faces challenges with illegal activities, which hinder its widespread adoption. To address this issue, this study focuses on detecting illicit activities on the Ethereum blockchain, particularly identifying illegal accounts using machine learning techniques. It proposes a novel method combining a convolutional neural network with an XGBoost classifier, aimed at differentiating between normal and illicit accounts based on transaction history. The XGBoost model, a tree-based ensemble classifier, not only improves accuracy but also prevents overfitting and enhances the model's generalizability. Its parallel tree-building feature also speeds up training, making the model more scalable. The model is trained on a balanced dataset of over 4,000 Ethereum accounts, encompassing both normal and fraudulent accounts. The results are promising, with the model achieving an accuracy of 99.7% and an average AUC of 0.9998 (std: 0.0008), outperforming standard machine learning models. This approach shows significant potential in combating illegal activities within the Ethereum blockchain.
Alok Pal Jain, K. Praveena, R J Anandhi, Sandeep Kumar · 6 authors
Blockchain and Machine Learning integration has changed the area of Regulatory Technology (RegTech), offering automated compliance solutions for difficult regulatory issues. By combining the benefits of blockchain technology with machine learning algorithms, this article introduces a unique Compliance Assurance Framework utilizing BlockchainEnhanced Machine Learning (CAF-BEML) to improve the effectiveness and trustworthiness of compliance procedures. The Blockchain-Enhanced Compliance Data Verification Algorithm (BCDVA), the Predictive Regulatory Risk Assessment Algorithm (PRRA), and the Secure Compliance Audit Trail Algorithm (SCATA) are the three central algorithms in the proposed framework, and they work together to guarantee data authenticity, forecast potential threats, and generate clear audit trails, all of which contribute to a robust regulatory ecosystem. The suggested method's effectiveness and superiority over current conventional approaches are proved, and its potential to change compliance management in numerous industries is emphasized via this in-depth study and comparison.
Ethereum and its native cryptocurrency, Ether, have played a worthy attention in the development of the blockchain and cryptocurrency space. Its programmability and smart contract capabilities have made it a foundational platform for decentralized applications and innovations across various industries. Because of its anonymous and decentralized structure, the hotheaded expansion of cryptocurrencies in the payment space has created both enormous potential and concerns related to cybercrime, including money laundering, financing terrorism, illegal and dangerous services. As more financial institutions attempt to integrate cryptocurrencies into their networks, there is an increasing need to create a more transparent network that can withstand these kinds of attacks. In this work, we are using different classification techniques, such as logistic regression (LR), random forest (RF), k-nearest neighbors (KNN), adaptive boosting (AdaBoost), and extreme gradient boosting (XGBoost) for Ethereum fraud detection. The dataset we are using includes rows of legitimate transactions done using the cryptocurrency Ethereum as well as known fraudulent transactions. The “XGBoost” model, which is noteworthy, detects variations that might attract notice and prevent potential issues in this chore.
Abstract: As a consequence of mass unemployment being the byproduct of COVID-19, people around the world discovered investment in cryptocurrency as a means to tackle their declining financial condition. Subsequently, the prominence of Ethereum as a platform for crypto transactions also gave rise to fraudulent transactions. The need to detect these frauds exists even today. This study proposes a token-based approach to detect fraud in Ethereum transactions incorporating the ERC20 standard, by employing machine learning techniques. After cleaning and preprocessing of the dataset, the transaction data was fed to Random Forest (RF), AdaBoost, Extra Trees (ET), Gradient Boosting (GB) and Extreme Gradient Boosting (XGB) classifiers in search of the most suitable model for fraud detection. Meticulous evaluation revealed that RF, ET and XGB classifiers yielded the highest accuracy of 95%. The proposed token-based approach hence presents a novel and efficient solution for fraud detection, with room for improvement and scalability.
Raghavendra Varma Cherukuri, G. Lavanya Devi, Neelapu Ramesh
Federated learning is a privacy-preserving machine learning technique that allows mutually distrusting parties to collaboratively train a model without sharing their data. Most federated learning techniques require a centralized aggregator that stores and aggregates models received from multiple parties. However, having a centralized entity may lead to a single point of failure problem. Another problem of federated learning is the leakage of sensitive data through model updates. To address these issues, we propose a Blockchain-based protocol for federated learning. Our protocol uses Blockchain as a model aggregator solving single-point-of-failure problems. Also, we use a Blockchain-based privacy-preserving technique to avoid data leakage problems. We also incorporate a Blockchain-based incentive distribution module to distribute incentives to model contributors. We perform experiments with well-known datasets and show that the proposed model's accuracy is close to that of a centralized aggregator. We also show the overhead of Blockchain by implementing the protocol and running it on Ethereum Blockchain.
Abstract: Fraudulent transactions significantly impact blockchain network trust and the economy. Traditional consensus methods (e.g., proof of work or proof of stake) can't confirm the identity of participants, leaving the network susceptible to fraud. Machine learning algorithms offer a potential solution to detect fraudulent transactions and participants. Fraudulent exchanges in the blockchain economy deter investors and raise skepticism. This study explores the effectiveness of controlled AI and deep learning models in identifying fraudulent transactions and users, integrating machine learning with blockchain technology.
Phishing scams have become one of the primary frauds on Ethereum, leading to substantial financial losses for users. Therefore, developing an effective phishing detection method for Ethereum holds significant importance for the blockchain ecosystem. However, the heterogeneity of Ethereum poses challenges when it comes to detecting phishing scams. In this paper, we propose a Curriculum Learning-based approach (CLPSD) for Ethereum phishing detection. We collect transaction records to build a graph. In order to distinguish the varying degrees of contribution among nodes, we design a difficulty measurer combining Local Outlier Factor and information entropy. Thus, CLPSD utilizes Graph Convolutional Network to initially learn from easy samples and progressively advance to more complex ones. The experimental results demonstrate that CLPSD outperforms existing methods, highlighting the superior performance of our approach and its significant relevance in enhancing the Ethereum ecosystem.
The increasing complexity of fraud in large enterprises, particularly in financial and transactional ecosystems, necessitates a scalable, agile, and distributed detection approach. Traditional centralized fraud detection architectures struggle to keep pace with the real-time requirements, data silos, and domain-specific fraud patterns that arise across organizational units. This paper explores the application of Data Mesh architecture as a decentralized and domain-oriented paradigm for enhancing fraud detection in large enterprises. Data Mesh shifts the ownership of data from centralized teams to domain-specific teams, treating data as a product and enabling better scalability, autonomy, and responsiveness. We propose a Data Mesh-based fraud detection model wherein each business domain—such as sales, finance, customer relations, and operations—operates as a semi-autonomous node capable of detecting fraud patterns locally while contributing to an enterprise-wide fraud intelligence network. Leveraging a federated governance model, the architecture facilitates standardized yet decentralized policy enforcement, model deployment, and cross-domain collaboration. This paper examines how data product thinking, domain-driven design, self-serve data platforms, and federated computational governance work together to create a resilient and adaptable architecture for fraud detection. The proposed methodology utilizes distributed anomaly detection algorithms, local event-driven stream processing (e.g., Apache Kafka and Flink), and inter-domain feedback loops for continuous model retraining and behavior correlation. Experimental simulations conducted on synthetic multi-domain enterprise data reveal improved time-to-detection, reduced false positives, and enhanced fraud detection in low-signal data scenarios compared to centralized models. Furthermore, the architecture demonstrates superior scalability and flexibility when integrating new domains and updating detection logic. The findings of this research indicate that Data Mesh not only democratizes access to fraud-related data but also enhances detection capabilities by aligning technical solutions with organizational complexity. This paper contributes to the growing body of decentralized AI applications in enterprises and offers actionable design patterns for implementing domain-centric fraud analytics in large organizations. Future work includes extending this architecture to incorporate privacy-preserving technologies such as federated learning and exploring its applicability in regulatory compliance frameworks
Rodrigo Colnago Contreras, Vitor Trevelin Xavier da Silva, Igor Trevelin Xavier da Silva, Monique Simplicio Viana · 8 authors
Since financial assets on stock exchanges were created, investors have sought to predict their future values. Currently, cryptocurrencies are also seen as assets. Machine learning is increasingly adopted to assist and automate investments. The main objective of this paper is to make daily predictions about the movement direction of financial time series through classification models, financial time series preprocessing methods, and feature selection with genetic algorithms. The target time series are Bitcoin, Ibovespa, and Vale. The methodology of this paper includes the following steps: collecting time series of financial assets; data preprocessing; feature selection with genetic algorithms; and the training and testing of machine learning models. The results were obtained by evaluating the models with the area under the ROC curve metric. For the best prediction models for Bitcoin, Ibovespa, and Vale, values of 0.61, 0.62, and 0.58 were obtained, respectively. In conclusion, the feature selection allowed the improvement of performance in most models, and the input series in the form of percentage variation obtained a good performance, although it was composed of fewer attributes in relation to the other sets tested.
Globally, increasing amount of cryptocurrency mining demand presents both opportunities and challenges for electric energy systems. This research employs a data-driven method to predict cryptocurrency mining load-tripping events, specifically targeting the low-voltage ride-through (LVRT) problem. The study utilizes diverse low-voltage fault scenarios generated through electromagnetic transient program (EMTP) software as training data. For fault classification, a convolutional neural network (CNN) is employed to improve model accuracy. Additionally, model explainability is enhanced using a decision tree for forecasting tripping events. The proposed approach is validated on a 6-bus power system integrated with cryptocurrency mining facilities.
In a Blockchain network the Certificate Authority (CA) is a vital component. The decentralization feature of the blockchain enables the network to have different CA for different transaction. The CA selection process is executed every time with each new transaction. This process is fine with small size network. However, if the network size increases this selection process can take numerous amounts of time, as the algorithm chosen to select the CA needs to search in entire network which increases the computational time and overhead of the network. If this search space can be decreases then, the resulting computational time and overhead can be decreased. The search space can be reduced by implementing the Agglomerative clustering technique. Each cluster is identified by a cluster index (CV) the average of the response time (R) and Validation Time (V). The nodes want to initiate transaction needs to select a appropriate cluster by entering a target budget (TB). The cluster having the CV less than TB is selected and then, Proof of Stake (PoS) consensus mechanism is applied to select the CA. Finally, the proposed model is evaluated over gas utilization, response time and validation time comparison.
Shimal Sh. Taher, Siddeeq Y. Ameen, Jihan A. Ahmed
In recent years, cryptocurrencies have experienced rapid growth and adoption, revolutionizing the financial sector. However, the rise of digital currencies has also led to an increase in fraudulent transactions and illegal activities. In this paper, we present a comprehensive study on the detection of fraudulent transactions in the context of cryptocurrency exchanges, with a primary focus on the Ethereum network. By employing various Machine Learning (ML) techniques and ensemble methods, including the hard voting ensemble model, which achieved a remarkable 99% accuracy, we aim to effectively identify suspicious transactions while maintaining high accuracy and precision. Additionally, we delve into the importance of eXplainable Artificial Intelligence (XAI) to enhance transparency, trust, and accountability in AI-based fraud detection systems. Our research contributes to the development of reliable and interpretable models that can significantly improve the cryptocurrency ecosystem security and integrity.
Yoo-Young Cheong, La Yeon Choi, Ji-Hwan Shin, Tae-Kyung Kim · 6 authors
Smart contracts are self-executing programs that are executed on blockchain platforms, and they have been widely used in recent years. However, malicious exploitation of the characteristics of smart contracts has become a pressing problem in blockchain security. Most of the existing methods have the drawback of detecting only a single type of vulnerability. To solve this problem, this study proposes a model for detecting multiple vulnerabilities in smart contracts. We preprocessed the data and transformed the Opcodes of smart contracts' source code into a control flow graph. We then extracted node features that are suitable to be the input of a graph neural network using Sent2Vec and performed graph classification. The proposed model was evaluated using real smart contracts, and the experimental results demonstrated that the proposed model can simultaneously detect multiple vulnerabilities with high performance.
Palarapu Saket, P. Jyothi, Arasada B Venkata Ayush Patnaik, Nagidi Chaithanya Vardhan Reddy · 5 authors
Ethereum has become one of the most popular blockchains in the world ever since its inception. There are now over 207 million Ethereum accounts and more than 6000 blocks are mined every day. Its fame also attracts various kinds of frauds so it's crucial to detect these frauds to keep Ethereum network sustainable and healthy. The main aim of the paper is to compare various fraud detection methods, besides trying to minimize false positives, and finally suggest the model best suitable for the task of fraud detection in Ethereum transactions by using various evaluation metrics.
Mohammad H. Hasan, Mohammad Shahriar Rahman, Helge Janicke, Iqbal H. Sarker
As the use of Blockchain for digital payments continues to rise in popularity, it also becomes susceptible to various malicious attacks. Successfully detecting anomalies within Blockchain transactions is essential for bolstering trust in digital payments. However, the task of anomaly detection in Blockchain transaction data is challenging due to the infrequent occurrence of illicit transactions. Although several studies have been conducted in the field, a limitation persists: the lack of explanations for the model's predictions. This study seeks to overcome this limitation by integrating eXplainable Artificial Intelligence (XAI) techniques and anomaly rules into tree-based ensemble classifiers for detecting anomalous Bitcoin transactions. The Shapley Additive exPlanation (SHAP) method is employed to measure the contribution of each feature, and it is compatible with ensemble models. Moreover, we present rules for interpreting whether a Bitcoin transaction is anomalous or not. Additionally, we have introduced an under-sampling algorithm named XGBCLUS, designed to balance anomalous and non-anomalous transaction data. This algorithm is compared against other commonly used under-sampling and over-sampling techniques. Finally, the outcomes of various tree-based single classifiers are compared with those of stacking and voting ensemble classifiers. Our experimental results demonstrate that: (i) XGBCLUS enhances TPR and ROC-AUC scores compared to state-of-the-art under-sampling and over-sampling techniques, and (ii) our proposed ensemble classifiers outperform traditional single tree-based machine learning classifiers in terms of accuracy, TPR, and FPR scores.
In this paper, we introduce the concept of Reputation Oracles: programs that can predict whether a given smart contract is reputable, or malicious. Our approach to classification combines state-of-the-art deep learning techniques with one of the least-explored types of onchain data: decompiled smart contract bytecode. To get a baseline for reputation, we trained our model on thousands of smart contracts representing not only reputable applications, such as leading DeFi protocols, but also malicious contracts that implement scam functionality like rug pulls, phishing, or spoofing. Remarkably, the resulting model boasts 94% test accuracy, outperforming a naive baseline model in specificity, negative predictive value, and overall accuracy. Under the hood, the model leverages transfer learning by applying the encoder portion of a transformer for feature representation, followed by a specifically trained feed-forward neural network (FFN) layer for the classification of smart contracts. Although initially applied to reputation prediction, our findings indicate that decompiled bytecode preserves enough relevant information from the original source code to be effectively used for various other classification tasks, thereby enabling better contextualization of smart contracts and the primitives they implement.
This study introduces an interpretable imbalanced data classification method for detecting cryptocurrency transaction fraud. We address data imbalance using SMOTE oversampling and data augmentation through contrastive learning. Next, we introduce a Transformer-based deep learning model that learns sample relevance. The model undergoes pre-training with a contrastive loss and fine-tuning through Bayesian optimization to effectively extract high-dimensional, higher-order, and fraud-related features. We employ a SHAP-based interpreter along with attention scores to elucidate the role of various transaction features in fraud detection. Comparative results demonstrate the model's remarkable recall performance in identifying cryptocurrency transaction fraud. Furthermore, it achieves an excellent F1 value, striking a balance between accuracy and recall. This research not only enriches financial fraud detection but also enhances cryptocurrency transaction security, promotes market development, and contributes to economic stability and social security.