Voice Assistance Privacy via Noise Learning Model
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing voice assistance systems lack effective methods for protecting user privacy during voice assistance services, particularly when providing personalized responses based on user context and intent, as they often transmit sensitive information directly to servers.
Innovation Solution
A system and method that generates second query information by adding noise information to first query information using a noise learning model, which is then transmitted to a server, allowing the device to remove noise-related responses and protect user privacy by not transmitting explicit user data.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If user context information is transmitted directly to the server for personalized voice assistance, then service personalization and accuracy are improved, but user privacy protection deteriorates
Solution Approach 1:
A noise learning model is introduced as an intermediary component between the user context information and the server. The model adds controlled noise to the query information, transforming sensitive user data into a protected format that preserves service accuracy while preventing direct exposure of personal information to the server.
Solution Approach 2:
The system changes the parameters of the query information by adding noise components generated through the noise learning model. This transformation modifies the data representation while maintaining the essential meaning needed for accurate voice assistance, thereby protecting privacy without sacrificing service quality.
2Object-affected harmful factors
If noise information is added to query information using a noise learning model, then user privacy is protected, but system complexity increases
Solution Approach 1:
The noise learning model operates autonomously within the device, generating and adding appropriate noise to query information without requiring external intervention or complex centralized processing. This self-service approach protects privacy while keeping the added complexity localized and manageable within the individual device.
3Object-affected harmful factors
If sensitive user information is not transmitted to the server, then privacy protection is improved, but service personalization capability deteriorates
Solution Approach 1:
The system creates a noisy copy of the user context information that preserves the essential patterns and meanings needed for personalized service while removing or obscuring directly identifiable sensitive data. This copied and transformed information is then transmitted to the server, enabling personalization without direct exposure of raw sensitive information.
Data Source
AI summary
An artificial intelligence (AI) system using a machine learning algorithm such as deep learning, and an application thereof are provided. A method of providing, by a device, a voice assistance service includes obtaining a voice input of a user, receiving certain context information from at least one peripheral device, generating first query information from the received context information and the voice input, generating second query information including noise information by inputting the first query information into a noise learning model, transmitting the generated second query information to a server, receiving, from the server, response information obtained based on the transmitted second query information, generating a response message by removing response information corresponding to the noise information from the received response information, and outputting the response message.