Probabilistic Transaction Model for Accurate Data Simulation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current data simulation technologies often fail to accurately simulate real-world transactions, inefficiently use processing resources, and may expose confidential information, as they generate irrelevant or dissimilar simulation data.
Innovation Solution
A transaction simulation platform using machine learning and probabilistic transaction models, such as Gaussian Mixture Models (GMM) or Generative Adversarial Networks (GAN), separates real transaction data into streams, trains models on sample distributions similar to actual ones, and generates simulated data using random numbers, conserving resources and protecting confidentiality.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If current data simulation technologies generate simulation data, then simulation data is produced, but the data is irrelevant or dissimilar to real-world transactions, reducing accuracy
Solution Approach 1:
The patent applies copying by training machine learning models on actual transaction data to learn the underlying statistical distributions and patterns. The model then generates simulated data that copies the characteristics and relationships of real transaction data, ensuring both accuracy and relevance without directly exposing confidential information.
Solution Approach 2:
The patent uses parameter changes by adjusting the statistical parameters (mean, variance, covariance) of the training data to generate simulated data with controlled characteristics. This allows the system to produce accurate simulations while maintaining the ability to modify data distributions as needed for different scenarios.
2Productivity
If traditional simulation methods are used, then simulation data is generated, but processing resources are used inefficiently due to error correction and re-generation needs
Solution Approach 1:
The patent implements feedback by using machine learning models that learn from actual transaction data and continuously improve their simulation accuracy. The model receives feedback during training on the statistical properties of real data and adjusts its parameters accordingly, reducing the need for error correction and re-generation in production.
Solution Approach 2:
The patent applies preliminary action by pre-training the simulation model on comprehensive transaction data before deployment. This preliminary training phase establishes accurate statistical distributions and patterns, so that when the model generates simulation data in production, it does so efficiently without requiring frequent corrections or re-generation.
3Measurement precision
If real transaction data is used for simulation, then accurate simulation data can be generated, but confidential information may be exposed
Solution Approach 1:
The patent applies copying by creating a statistical copy of transaction data through machine learning models rather than directly using or storing actual transaction records. The model captures the essential patterns and distributions of real data while generating synthetic samples that do not contain or reveal confidential information about specific transactions or customers.
Data Source
AI summary
A device may obtain, for a set of transactions, a set of transaction values associated with a particular industry. The device may determine one or more sample statistical distributions for a probabilistic transaction model by using one or more machine learning techniques. The one or more sample statistical distributions may be similar to one or more actual statistical distributions that are associated with the set of transaction values. The device may generate simulated transaction information using the probabilistic transaction model. The device may perform one or more actions after generating the simulated transaction information.