Behavior Classification Model Updating via Correction Data
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for determining the type of office applications, such as desktop cloud, voice conference, and video conference, are inefficient due to the need for manual labeling of training samples for machine learning algorithms.
Innovation Solution
A data stream classification model updating method that automatically determines the data stream class of a current data stream by using packet information and a behavior classification model, and updates the model based on correction data when the initial classification differs from the accurate classification.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If manual labeling of training samples is used to train machine learning models for data stream classification, then the model can be trained with accurate labeled data, but the process is low in efficiency and time-consuming
Solution Approach 1:
The system automatically generates correction data by comparing the classification results from the behavior classification model with the target correspondence, eliminating the need for manual labeling. The device self-updates its training samples by identifying and correcting misclassified data streams, thereby improving both efficiency and maintaining accuracy.
Solution Approach 2:
The system uses the target correspondence as a reference to evaluate the classification results from the behavior classification model. When discrepancies are found, the system generates correction data and uses it to update the model, creating a feedback loop that continuously improves classification accuracy without manual intervention.
2Reliability
If a behavior classification model is trained offline with pre-collected and manually labeled sample data, then the model structure can be established, but the model cannot adapt to new data stream types and requires frequent manual updates
Solution Approach 1:
The system transitions from a static offline-trained model to a dynamic online-updating model. The behavior classification model is continuously updated with correction data generated from actual data stream classification tasks, allowing it to adapt to new data stream types while maintaining its stable structure.
Solution Approach 2:
The system pre-establishes a target correspondence that maps common features to data stream classes. This preliminary structure serves as a reference framework that guides the online updating process, ensuring that new data is classified according to established patterns while allowing for adaptations.
3Measurement precision
If the behavior classification model is updated frequently with new training samples, then the classification accuracy improves, but the computational resources and time required for training increase
Solution Approach 1:
Instead of retraining the entire model with all historical data, the system updates the behavior classification model incrementally using only the correction data generated from recent classification tasks. This partial updating approach improves accuracy with minimal time and computational resources.
Solution Approach 2:
The system changes the parameters of the behavior classification model by incorporating correction data that highlights specific misclassification cases. By focusing on correcting erroneous classifications rather than retraining with all data, the model adapts efficiently with reduced computational overhead.
Data Source
AI summary
A data stream classification model updating method is disclosed. Determining, based on packet information of a current data stream and a behavior classification model, a first data stream class corresponding to the current data stream; determining, based on a target correspondence and a common feature of the current data stream, a second data stream class corresponding to the current data stream, where the target correspondence is a correspondence between a plurality of common features and a plurality of data stream classes; and if the first data stream class is different from the second data stream class, obtaining correction data corresponding to the current data stream, where the correction data includes the packet information of the current data stream and the second data stream class, and the correction data is used as a training sample to update the behavior classification model.