AI Training via Distributed Partial Data Blocks
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing AI training methods, such as federated learning, require significant computational capacity on user devices to process sensitive data locally, which is not feasible for devices like smartwatches or simple mobile phones, posing privacy concerns and resource constraints.
Innovation Solution
A system and method that distribute sensitive data into blocks, processed by multiple modules in an intermediate processing system, with each module generating partial models that are merged to form a global AI model, ensuring privacy protection without requiring extensive computational resources on source devices.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Object-affected harmful factors
If federated learning is used to process sensitive data locally on user devices, then privacy protection is improved, but computational resource requirements increase significantly
Solution Approach 1:
The patent segments sensitive data into multiple separate data blocks and distributes them to different processing modules. Each module processes only a portion of the data rather than the complete sensitive dataset, reducing the computational burden on individual devices while maintaining privacy protection through distributed processing.
2Object-affected harmful factors
If sensitive data is distributed to multiple processing modules, then privacy protection is improved, but system complexity increases
Solution Approach 1:
The patent introduces an intermediary system that coordinates between the data source device and multiple processing modules. This intermediary manages the distribution of data blocks, collects processing results, and orchestrates the training process, thereby managing system complexity centrally rather than requiring complex peer-to-peer coordination between all components.
3Productivity
If data is divided into blocks and processed in parallel, then training speed is improved, but coordination overhead increases
Solution Approach 1:
The patent implements continuous processing where multiple processing modules operate in parallel on different data blocks simultaneously, maintaining continuous useful action throughout the training process. The intermediary system manages this parallel processing continuously, collecting results as they become available and coordinating updates to the global model without interrupting the parallel computation flow.
Data Source
AI summary
Methods and systems to train artificial intelligence modules are described, which protect privacy of sensitive data by virtue of the fact that the data source extracts blocks of partial data from the source item of content and distributes the extracted partial data to a plurality of processing nodes in an intermediate processing system. The processing nodes each perform an initial portion of the training process, for example by performing a convolution of the partial data, to produce a partial model. The partial models are transmitted to a merging module which amalgamates them and completes the training process to generate a global model for the AI task.