Client-Server Neural Network Processing via Byte Array Transmission
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
High-load processing tasks, such as calculations using neural network models for interatomic potentials, require frequent and large-volume data exchanges between servers and clients, leading to prolonged communication times in SaaS systems.
Innovation Solution
Implementing an information processing system where the client transmits and the server process data as byte arrays without format conversion, reducing the need for serialization and deserialization, thereby shortening communication and processing times.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If data is transmitted with format conversion and serialization/deserialization, then data can be processed in different systems, but communication time and processing load increase
Solution Approach 1:
The patent applies preliminary action by converting data to byte arrays on the client side before transmission. This pre-processing ensures that data is already in the optimal format for server processing, eliminating the need for format conversion and serialization/deserialization operations on the server side, thus reducing communication time and processing load.
Solution Approach 2:
The patent extracts the data conversion and serialization operations from the server-side processing workflow and relocates them to the client side. By taking out these time-consuming operations from the critical communication path on the server, the system achieves faster processing while maintaining data compatibility through the use of universal byte array format.
2Measurement precision
If neural network calculations are performed with high-load processing, then accurate results are obtained, but processing time becomes excessively long
Solution Approach 1:
The patent applies preliminary action by preparing data in byte array format before neural network processing. This pre-prepared format allows the neural network to receive data immediately without waiting for format conversion or deserialization, thus maintaining calculation accuracy while significantly reducing the overall processing time.
3Adaptability or versatility
If data conversion and format transformation are performed, then data can be adapted to different processing requirements, but processing load increases
Solution Approach 1:
The patent extracts data conversion operations from the server-side processing pipeline and performs them on the client side before transmission. This extraction eliminates unnecessary format transformation steps on the server, maintaining data adaptability through client-side preprocessing while significantly improving processing efficiency by reducing redundant operations.
Solution Approach 2:
The patent implements self-service by enabling the client to autonomously convert data to byte arrays and prepare it for transmission. This self-service approach on the client side eliminates the need for server-side data conversion, maintaining data adaptability while improving overall system productivity by distributing the conversion workload to the client.
Data Source
AI summary
An information processing system includes a first information processing device and a second information processing device. The first information processing device is configured to receive the atomic information from the second information processing device, calculate a processing result corresponding to the atomic information by inputting the atomic information into a neural network, and transmit the processing result to the second information processing device. The second information processing device is configured to transmit atomic information to the first information processing device.