Knowledge Tracking Model Training with Data Augmentation for Overfitting
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Deep learning-based knowledge tracking models suffer from overfitting due to insufficient training data, leading to poor prediction accuracy in online education environments, particularly in domains lacking problem-solving data.
Innovation Solution
Implement data augmentation techniques such as replacing, inserting, or deleting problem-solving records to generate new training data, and apply a regularization loss function based on the augmentation method to optimize the model's performance.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If data augmentation is performed by replacing, inserting, or deleting problem-solving records, then the training data volume increases and overfitting is reduced, but the device complexity and processing time increase
Solution Approach 1:
The patent applies preliminary action by pre-defining multiple types of data augmentation operations (replacement, insertion, deletion) and their corresponding regularization loss functions before training. The system prepares augmentation templates and operation sets in advance, which are then applied systematically during the training process, reducing the need for complex real-time decision-making while maintaining effectiveness in reducing overfitting.
Solution Approach 2:
The patent implements parameter changes by systematically varying augmentation parameters such as the probability of replacement operations, the number of records to insert or delete, and the selection criteria for augmentation targets. These parameter adjustments allow the system to control the degree of data transformation and corresponding regularization strength, balancing the trade-off between reducing overfitting and managing processing complexity.
2Quantity of substance
If data augmentation is performed to generate new training data, then the training data volume increases, but the loss of information and data quality may increase
Solution Approach 1:
The patent applies feedback mechanisms by implementing regularization loss functions that continuously monitor and evaluate the quality of augmented data during training. The system compares predictions on original versus augmented data, uses this feedback to adjust augmentation parameters in real-time, and ensures that data quality degradation is detected and corrected, thereby maintaining high-quality training data while increasing volume.
Solution Approach 2:
The patent converts the potential harm of data quality loss into a benefit by using the regularization process itself to identify and correct quality issues. The discrepancy between original and augmented data predictions serves as a diagnostic tool, allowing the system to selectively refine or reject augmented samples that degrade quality, thereby transforming the information loss risk into a quality control mechanism.
Data Source
AI summary
Provided is an apparatus for training a knowledge tracking model, which is an apparatus for predicting a correct answer probability of a user on the basis of data augmentation, the apparatus including: a problem-solving data storage unit configured to store problem-solving data in which a problem solved by a user and a response of the user to the problem are mapped; a data augmentation performing unit configured to receive the problem-solving data from the problem-solving storage unit and convert the problem-solving data to generate augmented data; a regularization performing unit configured to receive the augmented data from the data augmentation performing unit and perform a regularization operation using a regularization loss function determined on the basis of a data augmentation method that is performed; and a model training unit configured to input the augmented data to a knowledge tracking model, allow the knowledge tracking model to learn a weight representing a relationship between a problem-solving record included in the augmented data and a correct answer probability of the user, and train the knowledge tracking model for predicting a correct answer probability of the user for a specific problem on the basis of the weight.