ML Pre-Processor Configuration Adjustment
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional machine learning systems lack the ability to dynamically adjust the configuration of pre-processors and post-processors based on the outputs of the machine learning model, leading to inefficiencies in resource usage and increased latency.
Innovation Solution
A system and method for dynamically adjusting the configuration of pre-processors and post-processors in a machine learning system by using a controller to adjust settings based on inference data and statistical data, including memory usage, workload, and resource usage, with adjustments made periodically or based on metadata and user inputs.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If the configuration of pre-processor and post-processor is fixed in conventional machine learning systems, then the system structure is simple and easy to implement, but resource usage efficiency is low and latency is increased
Solution Approach 1:
The patent applies dynamics by making the pre-processor and post-processor configurations adjustable rather than fixed. The controller dynamically modifies configuration parameters of these processors based on real-time statistical data from the machine learning model outputs, enabling the system to adapt its processing capabilities to match actual workload requirements, thereby improving resource usage efficiency without requiring a completely complex restructured system
Solution Approach 2:
The patent implements parameter changes by modifying configuration parameters of the pre-processor and post-processor based on statistical data. The controller adjusts specific configuration parameters (such as processing thresholds, data sampling rates, or transformation intensities) according to the characteristics of model outputs, allowing the system to optimize resource utilization dynamically while maintaining a relatively simple overall architecture
2Loss of time
If the configuration of pre-processor and post-processor is fixed, then the system is easier to operate and maintain, but latency is increased due to inability to optimize processing based on actual data characteristics
Solution Approach 1:
The patent applies feedback by using statistical data from machine learning model outputs to inform configuration adjustments of the pre-processor and post-processor. The controller continuously monitors processing characteristics and uses this feedback to dynamically adjust configurations, reducing latency by optimizing processing parameters based on actual data patterns while automating the complexity of configuration management
Solution Approach 2:
The patent implements self-service by enabling the system to automatically adjust its own configuration without requiring manual intervention. The controller autonomously modifies pre-processor and post-processor settings based on statistical data, allowing the system to self-optimize processing latency while reducing the operational burden on users
Data Source
AI summary
Systems, methods, and computer-readable media are disclosed for dynamically adjusting a configuration of a pre-processor and/or a post-processor of a machine learning system. In one aspect, a machine learning system can receive raw data at a pre-processor where the pre-processor being configured to generate pre-processed data, train a machine learning model based on the pre-processed data to generate output data, process the output data at a post-processor to generate inference data, and adjust, by a controller, configuration of one or a combination of the pre-processor and the post-processor based on the inference data.