Encrypted AI Model Explanation System
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
When AI or ML models are accessed over a cloud network, data security and privacy are compromised due to the exposure of actual data, and encrypted outputs lack adequate explanation, making it difficult for users to understand the reasoning behind the AI's decisions.
Innovation Solution
A method and system that encrypts both input data and AI models using public encryption keys, generates encrypted outputs and explanations, and decrypts them at the client device using private keys, ensuring data security and providing explainable AI outputs.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If data encryption is applied to protect data security and privacy when using AI models in cloud network, then data security and privacy are improved, but the output cannot be adequately explained and users cannot understand the reasoning
Solution Approach 1:
The system segments the explanation generation process from the main AI model processing. It separates the encrypted model into an explanation generation component that produces encrypted explanations, which are then decrypted and rendered separately at the client device. This allows the model to operate in encrypted form for security while still providing understandable explanations through a dedicated explanation pathway.
Solution Approach 2:
The system introduces an intermediary explanation rendering component that acts as a bridge between the encrypted AI model output and the user. This intermediary receives encrypted explanations from the model, decrypts them using the private key, and renders them in a human-understandable format. This mediator enables both encrypted secure processing and readable explanation delivery without exposing raw data.
2Manufacturing precision
If actual data is exposed to train AI models or generate output in cloud network, then model accuracy and functionality are improved, but data privacy and security are compromised
Solution Approach 1:
The system changes the state parameter of the data from plaintext to encrypted form throughout the AI model processing pipeline. By transforming data into encrypted parameters that can still be processed by the model (through homomorphic encryption or similar techniques), the system maintains model functionality and accuracy while preventing data exposure. The model operates on encrypted data parameters without needing to decrypt them.
Solution Approach 2:
The system extracts the decryption capability from the cloud-based AI model and relocates it to the client device. The model processes data in encrypted form in the cloud, but the actual decryption and explanation rendering occur locally on the user's device. This extraction ensures that sensitive data never exposes the cloud environment while maintaining model accuracy through proper cryptographic processing.
Data Source
AI summary
This disclosure relates to method and system for providing explanation for output generated by artificial intelligence (AI) model. The method may include receiving encrypted input data and a public encryption key from a client device, wherein the encrypted input data is encrypted using the public encryption key. The method may further include generating an encrypted AI model by encrypting an AI model using the public encryption key. The method may further include generating an encrypted output and an encrypted feature data based on the encrypted input data using the encrypted AI model, and generating an encrypted explanation for the encrypted output based on the encrypted feature data. The method may further include providing the encrypted output and the encrypted explanation to the client device for rendering, wherein the encrypted output and the encrypted explanation are decrypted by the client device using a private encryption key.