Blockchain Fraud Prediction via Transaction Data Aggregation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for identifying fraudulent transactions in blockchain-based cryptocurrencies are cumbersome and computationally expensive due to the large number of inputs and outputs in each transaction, making real-time analysis challenging.
Innovation Solution
A machine learning-based method that aggregates input and output data from blockchain transactions into a manageable dataset, allowing for the training of a model to predict fraudulent transactions efficiently and accurately.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If heuristic methods are used to analyze each input and output addressee for fraudulent activity, then fraud detection accuracy is improved, but computational cost increases exponentially
Solution Approach 1:
The patent segments the analysis by dividing transactions into input and output components, analyzing them separately rather than as a monolithic unit. This allows the system to process each component independently, reducing the computational complexity from analyzing all combinations to analyzing individual segments.
Solution Approach 2:
The patent applies partial action by analyzing only the most relevant features of inputs and outputs (such as address characteristics, transaction patterns, and network relationships) rather than performing exhaustive analysis of all possible data points. This selective approach maintains detection accuracy while significantly reducing computational burden.
2Reliability
If all input and output data rows are processed individually, then detection thoroughness is improved, but processing time increases
Solution Approach 1:
The patent merges the analysis of multiple input and output data rows by aggregating their characteristics into combined feature sets. Instead of processing each row completely separately and then combining results, the system integrates the analysis at the feature level, processing aggregated data that captures the essential properties of all inputs and outputs simultaneously.
Solution Approach 2:
The patent performs preliminary aggregation of input and output data before the main fraud detection analysis. By pre-processing and combining data rows into aggregated representations that capture key characteristics, the system reduces the volume of data requiring detailed analysis while preserving the information necessary for thorough detection.
3Measurement precision
If feature-rich transaction data is analyzed in real-time, then prediction accuracy is improved, but system complexity increases
Solution Approach 1:
The patent extracts and isolates the most discriminative features from transaction data that are most indicative of fraudulent activity. By identifying and extracting only the critical features (such as address reputation scores, transaction pattern anomalies, and network relationship indicators) rather than processing all available features, the system maintains high prediction accuracy while reducing computational complexity.
4Reliability
If comprehensive transaction analysis is performed, then fraud detection reliability is improved, but real-time processing capability deteriorates
Solution Approach 1:
The patent applies partial action by performing comprehensive analysis only on the most critical and informative features of transactions, while using simplified analysis for less critical aspects. This selective comprehensive approach maintains detection reliability for the most important fraud indicators while enabling real-time processing by avoiding unnecessary computational overhead on less significant data elements.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
A computer implemented method of training a model, using a machine learning process, to predict whether a transaction of a digital currency stored in a blockchain is fraudulent, comprises: unpacking (202) a block in the blockchain into a table comprising one or more rows of input and output data for a previous transaction stored in the block and aggregating (204) the one or more rows of input and output data to form an aggregated row of transaction data for the previous transaction. The method further comprises labelling (206) the aggregated row of transaction data for the previous transaction according to whether the previous transaction was fraudulent and using (208) the aggregated row of transaction data and the label as training data with which to train the model.