Federated Customer Experience Modeling Without Raw Data Sharing
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing customer experience management systems face challenges in securely and accurately modeling customer experience perception due to data silos, privacy concerns, and the difficulty in aggregating fragmented data across different parties, leading to inaccurate and incomplete models.
Innovation Solution
Implementing federated learning to enable secure, joint customer experience perception modeling and evaluation by allowing parties to perform local model training and share only intermediate results, maintaining data privacy and security while enhancing model accuracy and comprehensiveness.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If joint aggregation of customer data is performed to improve model accuracy, then model precision is improved, but data privacy is compromised
Solution Approach 1:
The patent segments the centralized model training process into distributed local training across multiple parties. Each party trains models locally on their own data without sharing raw data, thus maintaining privacy while collectively improving model accuracy through federated aggregation of model parameters or gradients.
Solution Approach 2:
The patent introduces a federated learning coordinator as an intermediary that manages the collaborative training process without accessing actual customer data. The coordinator aggregates model updates from participating parties and distributes global model parameters, enabling accurate joint modeling while preserving data privacy through encrypted or differential privacy-protected communications.
2Reliability
If data is kept in separate silos to maintain privacy and security, then data security is improved, but model comprehensiveness deteriorates
Solution Approach 1:
The patent merges the capabilities of multiple distributed data silos through federated learning, where model knowledge from each party is combined into a comprehensive global model. This allows the system to leverage diverse data sources across parties while keeping raw data localized, thus maintaining security while achieving comprehensive modeling coverage.
Solution Approach 2:
The patent transitions the collaboration from the data dimension (sharing raw customer data) to the model parameter dimension (sharing trained model weights and gradients). This dimensional shift enables parties to contribute their data value without exposing sensitive information, achieving both security and comprehensiveness simultaneously.
3Measurement precision
If centralized data aggregation is performed to improve model accuracy, then model precision is improved, but system complexity increases
Solution Approach 1:
The patent implements self-service local training at each party's endpoint, where models are trained autonomously on local data without requiring centralized data collection infrastructure. This distributes the computational workload and reduces the complexity of data aggregation systems while maintaining model accuracy through coordinated federated learning updates.
Data Source
AI summary
The present disclosure relates to customer experience perception based on federated learning. An electronic device for a service provider in a wireless communication system, the service provider capable of implementing customer experience perception with at least one other service provider in the wireless communication system through federated learning, the electronic device comprises a processing circuit configured to: determine training data for training a global model related to the customer experience perception based on the federated learning, wherein the training data is related to the customer's experience for a specific service/product/business; perform local model training by using the determined training data to obtain intermediate information related to local model training; and transmit the intermediate information to a coordinator, so that the coordinator can train a customer experience perception model by aggregating the intermediate information from the provider and at least one other provider.