Neural Network Scoring Optimization via Outlier Neuron Reassignment
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing neural network models, particularly ensemble models, face challenges in achieving real-time scoring performance due to inefficiencies in processing time, which can be attributed to outlier neurons impeding rapid processing within neuron clusters.
Innovation Solution
The implementation of a multi-layered virtual computation module that identifies outlier neurons within clusters, reassesses their association, and adjusts cross-layer functionality based on dependencies, thereby optimizing neural network performance in real-time.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Speed
If traditional neural network processing is used, then model accuracy is maintained, but processing speed is insufficient for real-time scoring
Solution Approach 1:
The patent segments the neural network into multiple layers with clustered neurons, organizing them into a multi-layered virtual computation module. This segmentation allows parallel processing across layers and clusters, significantly improving processing speed while maintaining the ability to perform real-time scoring operations.
Solution Approach 2:
The patent introduces a new dimensional organization by creating a multi-layered virtual computation module structure that adds hierarchical depth to the neural network. Neurons are organized into clusters within layers, which are then arranged in multiple computational layers, creating a multi-dimensional processing architecture that enhances throughput for real-time operations.
2Productivity
If outlier neurons are left in their original clusters, then cluster structure is simple, but processing bottlenecks occur
Solution Approach 1:
The patent identifies and extracts outlier neurons from their original clusters based on computational characteristics. These outlier neurons are then reassign ed to different clusters where their computational patterns are more appropriate, eliminating processing bottlenecks while maintaining overall system efficiency.
Solution Approach 2:
The patent applies local optimization by reassigning only the outlier neurons that cause processing bottlenecks, rather than restructuring entire clusters. This localized approach improves processing efficiency in specific areas while minimizing the overall complexity of cluster reassignment.
3Adaptability or versatility
If cross-layer dependencies are not adjusted, then layer structure is stable, but workload optimization is limited
Solution Approach 1:
The patent implements dynamic adjustment of cross-layer functionality based on workload characteristics. The multi-layered virtual computation module can adapt its cross-layer dependencies and neuron assignments in real-time according to the specific computational workload, optimizing performance for different task types while managing adjustment complexity through systematic approaches.
Data Source
AI summary
Systems, computer program products and/or computer-implemented methods described herein relate to a process to optimize performance of an operating neural network. A system can comprise a memory that stores computer executable components and a processor that executes the computer executable components, which can comprise an identification component that, employing an operating, multi-layered virtual computation module of looped neurons, identifies a first neuron of a first cluster of a first layer of the looped neurons as being an outlier neuron, an adjustment component that reassigns the outlier neuron from the first cluster to a second cluster of the first layer, and a scheduling component that, based on a dependency among layers of the multi-layered virtual computation module, including the first layer, adjusts a cross-layer functionality of the looped neurons for a workload currently being performed by the multi-layered virtual computation module.