Machine-Learned Model Block Selection for Edge Resource Scaling
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
On-device machine-learned models face challenges with varying computational resources, leading to poor performance, increased latency, and suboptimal resource allocation due to fluctuating device capabilities.
Innovation Solution
A machine-learned model with adjustable computational demand is implemented, allowing for selective deactivation of blocks based on resource allocation parameters, using residual connections to maintain model functionality and quality.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If the full machine-learned model is executed on edge devices, then model accuracy and completeness are maintained, but computational resource consumption increases and processing time increases
Solution Approach 1:
The machine-learned model is divided into multiple independently executable blocks or layers. The system can selectively activate only the necessary subset of blocks based on resource allocation parameters, allowing the model to be segmented into optional computational units that can be dynamically enabled or disabled to balance accuracy and resource usage.
2Productivity
If more computational resources are allocated to the machine-learned model, then processing speed and performance improve, but device resource availability decreases for other tasks
Solution Approach 1:
The system dynamically adjusts the computational configuration of the machine-learned model based on real-time resource allocation parameters. The resource allocation parameter can be modified at runtime to enable or disable specific model blocks, allowing the system to adapt processing speed and resource consumption to current device conditions and competing task requirements.
3Loss of energy
If the machine-learned model is optimized for resource-constrained environments, then energy consumption and computational demand are reduced, but model performance and accuracy deteriorate
Solution Approach 1:
Different blocks within the machine-learned model can be selectively activated based on local conditions and resource allocation parameters. The system maintains high performance by activating only the most critical or performance-sensitive blocks when resources are limited, while less critical blocks can be deactivated to reduce energy consumption, creating local optimization rather than uniform model execution.
4Use of energy by moving object
If blocks are deactivated to reduce computational demand, then resource consumption decreases, but model robustness and output quality may be compromised
Solution Approach 1:
The system incorporates feedback mechanisms to monitor the impact of block deactivation on model output quality and performance. Based on this feedback and resource allocation parameters, the system can dynamically adjust which blocks remain active, ensuring that deactivation decisions maintain model robustness within acceptable thresholds while achieving resource consumption reduction goals.
Data Source
AI summary
An example computing device can store a plurality of blocks of a machine-learned model. When different subsets of the plurality of blocks of the machine-learned model are deactivated, different remaining subsets of the plurality of blocks are connectable to form different machine-learned models. The computing device can be configured to form an adapted machine-learned model of the different machine-learned models, the adapted machine-learned model comprising a remaining subset of the plurality of blocks that does not include a deactivated subset of the plurality of blocks. The computing device can be configured to, after forming the adapted machine-learned model, input a model input into the adapted machine-learned model to process the model input using the remaining subset of the plurality of blocks. The computing device can be configured to receive, from the adapted machine-learned model, a model output based on the model input.