Pseudo-Log Generation Using Generator-Discriminator Feedback
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing log generation methods face challenges in expanding log data sets, particularly when the number of real logs is insufficient, leading to deviations and inaccuracies in log analysis and troubleshooting.
Innovation Solution
A method involving a generator and discriminator to create pseudo-logs based on generating conditions and random noise, determining event types and authenticity, and training the generator and discriminator to improve the accuracy of pseudo-log generation, expanding the log data set effectively.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Quantity of substance
If pseudo-logs are generated using traditional methods, then the log volume can be expanded, but the accuracy and authenticity of the generated logs deteriorates
Solution Approach 1:
The patent implements a feedback mechanism where the discriminator evaluates the authenticity of generated pseudo-logs and provides feedback to the generator. The discriminator analyzes feature matrices of both real and pseudo-logs, determining authenticity scores that are used to update the generator's parameters through backpropagation, continuously improving the quality of generated logs
Solution Approach 2:
The patent introduces a discriminator as an intermediary component between the generator and the evaluation process. This discriminator acts as a mediator that assesses the authenticity of pseudo-logs by comparing their feature matrices against real logs, enabling indirect optimization of the generator without requiring direct access to ground truth log distributions
2Ease of manufacture
If a small number of real logs are used, then the data collection process is simplified, but the reliability of log analysis deteriorates
Solution Approach 1:
The patent creates copies of real log characteristics by training the generator on available real logs to learn their underlying distributions and patterns. The generator then produces multiple pseudo-log copies that replicate the statistical properties and event sequences of real logs, expanding the dataset while preserving analytical reliability
Solution Approach 2:
The patent transforms the limited real log data into a larger dataset by changing the parameter space through the generator-discriminator system. The generator learns parameters representing log distributions and generates diverse pseudo-logs by sampling different noise vectors and conditioning on various event types, effectively expanding the data parameter space
3Measurement precision
If more real logs are collected, then the accuracy of log analysis improves, but the time and resources required for data collection increases
Solution Approach 1:
The patent performs preliminary action by training the generator-discriminator system in advance using a small set of real logs. Once trained, the generator can rapidly produce large volumes of pseudo-logs without requiring additional real log collection, pre-computing the log distribution characteristics that would otherwise require extensive real data collection
Data Source
AI summary
Embodiments of the present disclosure provide a method, an electronic device, and a computer program product for generating logs. The method comprises: generating, by a generator, a pseudo-log based on a generating condition and random noise; determining, by a discriminator, event types of log messages in a real log and the pseudo-log; determining, by the discriminator, a feature matrix of log sequences in the real log and the pseudo-log based on the event types; determining, by the discriminator, authenticity of the real log and the pseudo-log based on the feature matrix; and training the generator and the discriminator based on the authenticity. The method according to embodiments of the present disclosure may improve the accuracy of a training model for generating pseudo-logs that are sufficiently real.