Encrypted Prediction System Using Abnormality Clusters
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current predictive analytics tools face challenges in securely computing predictions over networks, as they require raw data that can lead to privacy breaches and biased results due to the need for contextual information, and they lack efficiency in aggregating data from multiple sources.
Innovation Solution
The method involves computing predictions using entirely encrypted data, creating abnormality clusters that store indications of entities rather than raw data, allowing secure and unbiased predictions by analyzing queries based on these clusters, which are computed from aggregated encrypted datasets without revealing identifiable information.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If raw data is used for predictive analytics, then prediction accuracy can be improved, but privacy security deteriorates due to exposure of identifiable information
Solution Approach 1:
The patent introduces encrypted data as an intermediary between raw data and the predictive analytics system. The encryption acts as a mediator that allows the system to process data for accurate predictions while preventing direct access to identifiable information. The encrypted dataset serves as a buffer layer that maintains prediction capability while protecting privacy.
Solution Approach 2:
The patent creates an encrypted copy of the raw dataset that preserves the statistical properties and patterns needed for prediction while removing identifiable information. This encrypted copy can be shared and processed without exposing the original sensitive data, allowing accurate predictions to be made on replicated data structures.
2Reliability
If contextual information is collected for predictions, then prediction reliability improves, but data bias increases due to selective information gathering
Solution Approach 1:
The patent combines data from multiple sources into a single encrypted dataset, merging diverse information streams while maintaining encryption throughout the process. This consolidation allows the system to leverage multiple data sources for more reliable predictions while the encryption ensures that contextual information is aggregated without introducing bias through selective collection or processing.
3Measurement precision
If data from multiple sources is aggregated, then prediction accuracy improves, but network security complexity increases
Solution Approach 1:
The patent applies encryption to data before it leaves the source systems, performing the security protection action in advance. This preliminary encryption simplifies the overall network security architecture by eliminating the need for complex secure transmission and storage mechanisms at intermediate nodes, as the data is already protected before aggregation begins.
Data Source
AI summary
There is provided a method for computing an encrypted prediction in response to an encrypted query, comprising: obtaining an encrypted dataset comprising encrypted records for respective encrypted entities, each record storing encrypted parameter values of parameters and an associated indication of the respective entity, computing abnormality clusters according to the records of the encrypted dataset, wherein each of the abnormality clusters stores indications of entities of records of the encrypted dataset having mathematically significant common abnormal feature(s) that statistically differentiates records of the respective abnormality cluster from other records of the encrypted dataset, receiving a query comprising target indications of respective target entities associated with common feature(s), and analyzing the query according to the abnormality clusters to identify at least one encrypted result entity indication according to a likelihood of the encrypted result entity indication predicted to correlate to the common feature(s) at a future time interval.