AI Blockchain Pattern Detection with Missing Data Compensation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The detection of illegal, malicious, and fraudulent blockchain activities is challenging due to the immutable nature of blockchain records and the lack of high-quality training data for artificial intelligence models, which requires immediate and precise action to prevent fraud.
Innovation Solution
The system uses artificial intelligence models to identify patterns in blockchain activities based on multi-modal data, compensating for missing data points by training on digital asset proportions at subsets of blockchain accounts, allowing for real-time monitoring and visualization of relevant activities without generating individual intermediary activity data.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Difficulty of detecting and measuring
If artificial intelligence models are used to detect blockchain activities, then detection capability is improved, but the requirement for high-quality training data creates a bottleneck
Solution Approach 1:
The patent segments the blockchain network into monitored accounts and unmonitored accounts, and divides training data into sampled intermediary activities and unsampled intermediary activities. This segmentation allows the system to work with limited high-quality training data while still achieving effective detection by processing different types of data through different pathways in the model architecture.
Solution Approach 2:
The system applies partial action by sampling only a subset of intermediary blockchain activities for detailed analysis and model training, rather than processing all intermediary activities. This partial processing approach enables the system to function effectively with limited training data while maintaining detection capability through the use of sampled representative data.
2Loss of information
If all intermediary blockchain activities are processed, then completeness of analysis is improved, but computational complexity and time consumption increase
Solution Approach 1:
The patent extracts and processes only the essential features from intermediary blockchain activities - specifically the digital asset proportions at monitored accounts - rather than analyzing all raw transaction data. This extraction approach maintains the ability to detect fraud patterns while significantly reducing processing time and computational resources required.
Solution Approach 2:
The system performs partial processing by focusing computational resources on analyzing digital asset proportions at monitored accounts rather than examining every intermediary activity in detail. This selective approach preserves detection effectiveness while reducing overall processing time and resource consumption.
3Productivity
If digital asset proportions at monitored accounts are used for training, then training efficiency is improved, but data representativeness may be compromised
Solution Approach 1:
The patent applies local quality by using different data characteristics for different purposes: digital asset proportions at monitored accounts are used for efficient model training, while the same proportions are used to detect actual fraud patterns. This localized use of data quality characteristics allows the system to achieve both training efficiency and detection accuracy without compromising representativeness.
Data Source
AI summary
Systems and methods for identifying patterns in blockchain activities based on multi-modal data using artificial intelligence models that compensate for training data featuring a high proportion of missing data points. For example, the system may receive blockchain activity record data for a plurality of blockchain activities involving a plurality of blockchain accounts. The system may input the data into an artificial intelligence model, wherein the artificial intelligence model is trained to identify serial relationships of related blockchain activities corresponding to inputted target blockchain activities based on proportions of digital assets at subsets of blockchain accounts of the plurality of blockchain accounts. The system may receive an output from the artificial intelligence model. The system may generate for display, in a user interface, a visualization of the target blockchain activity based on the output.