Personalized User Classification via Local Model Adaptation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing user classification methods for terminal devices are ineffective due to the use of a common user classification model for all users, which fails to account for individual usage habits, leading to poor accuracy and security risks.
Innovation Solution
A method where a target terminal device obtains an initial user classification model from a server, updates it based on the user's operation data, and uses the updated model for personalized user classification, allowing for accurate classification of the user's identity.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If a common user classification model is used for all terminal devices, then the device complexity is reduced and ease of manufacture is improved, but the measurement precision of user classification and reliability of security are worsened
Solution Approach 1:
The user classification model is segmented into two parts: a common initial model deployed to all terminal devices, and a personalized model updated locally on each device based on individual user operation data. This segmentation allows the system to maintain low initial complexity while achieving high classification accuracy through personalized adaptation.
Solution Approach 2:
An initial user classification model is pre-trained and deployed to all terminal devices before use. This preliminary action provides a baseline classification capability that works for all users, and subsequent personalized updates refine this initial model based on individual user behavior patterns.
2Ease of operation
If a common user classification model is used for all users, then the ease of operation is improved and device complexity is reduced, but the reliability of user identification and security is worsened
Solution Approach 1:
The terminal device automatically updates its user classification model based on its own collected operation data without requiring manual intervention. The system performs self-learning and adaptation, maintaining ease of operation while improving reliability through personalized model refinement over time.
Solution Approach 2:
The system continuously collects user operation data and uses this feedback to update and refine the personalized user classification model. This feedback loop enables the model to adapt to individual user behavior patterns, improving identification accuracy while maintaining operational simplicity.
3Measurement precision
If personalized user classification is implemented for each user, then the measurement precision of user classification is improved, but the device complexity and loss of time for model training increase
Solution Approach 1:
A common initial user classification model is pre-trained and deployed to all terminal devices before personalized use. This preliminary action eliminates the need to train from scratch for each user, significantly reducing the time loss while still enabling subsequent personalized refinement for high accuracy.
Solution Approach 2:
The system updates the user classification model by adjusting parameters based on individual user operation data rather than retraining the entire model. This parameter change approach maintains high personalized accuracy while minimizing the time and computational resources required for adaptation.
Data Source
AI summary
A computer-implemented user classification method includes: obtaining, by a target terminal device, an initial user classification model from a server, in which the initial user classification model is provided by the server to multiple terminal devices, the multiple terminal devices including the target terminal device; obtaining first operation data of a registered user of the target terminal device; updating the initial user classification model based on the first operation data, to obtain an updated user classification model that is personalized for the registered user; and classifying, based on the updated user classification model, an identity of a current user of the target terminal device.