Adaptive Path Manager for Dynamic ANN Resource Allocation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing artificial neural networks (ANNs) face challenges in dynamically managing computing paths across heterogeneous resources, leading to inefficiencies in resource utilization and performance, particularly in adapting to changing operating environments and priorities.
Innovation Solution
A method and system that utilize adaptive path managers to initialize resource information, preference level metrics, and initial computing paths, dynamically setting computing paths based on resource and operating environments to optimize resource usage and adapt to changing conditions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of manufacture
If fixed computing paths are used in ANN, then implementation simplicity is maintained, but resource utilization efficiency deteriorates
Solution Approach 1:
The patent implements dynamic computing path selection by introducing a path manager that dynamically determines which computing path to use based on real-time resource availability and operating conditions. The system maintains multiple pre-defined computing paths (first computing path using first plurality of resources, second computing path using second plurality of resources) and dynamically switches between them, transforming the static system into a dynamic one that adapts to changing conditions while maintaining implementation simplicity through pre-defined path structures.
2Adaptability or versatility
If multiple heterogeneous resources are used to drive ANN, then computing capability and adaptability are improved, but path management complexity increases
Solution Approach 1:
The patent introduces a path manager as an intermediary component that handles the complexity of managing multiple heterogeneous resources. The path manager receives operation requests, determines the appropriate computing path based on resource availability and operating conditions, and routes operations to the suitable resources. This intermediary abstracts the complex resource management logic from the main ANN processing, thereby managing path management complexity while maintaining high computing capability and adaptability.
3Productivity
If dynamic path selection is implemented, then resource allocation optimization is achieved, but system complexity increases
Solution Approach 1:
The patent applies preliminary action by pre-defining multiple computing paths with specific resource allocations before runtime. The first computing path and second computing path are both established in advance, with their respective resource requirements and characteristics predetermined. When the ANN needs to perform operations, the path manager simply selects from these pre-configured paths based on current resource availability, avoiding the need for complex real-time path creation and optimization logic, thus achieving resource allocation optimization with controlled system complexity.
Data Source
AI summary
In a method of managing a plurality of computing paths in an artificial neural network (ANN) driven by a plurality of heterogeneous resources, resource information, preference level metrics, and a plurality of initial computing paths are obtained by performing an initialization. The resource information represents information associated with the heterogeneous resources. The preference level metrics represent a relationship between the heterogeneous resources and a plurality of operations. The initial computing paths represent computing paths predetermined for the operations. When a first event including at least one of the plurality of operations is to be performed, a first computing path for the first event is set based on the initial computing paths, the preference level metrics, resource environment, and operating environment. The resource environment represents whether the heterogeneous resources are available. The operating environment represents contexts of the ANN and at least one electronic device including the heterogeneous resources.