Merged Neural Network Layer for Computational Efficiency
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Processing sensor-originated data, such as image or audio data, with neural network systems is computationally intensive, necessitating improvements in computational efficiency.
Innovation Solution
A neural network system with a merged layer that combines weights from two neural networks to generate intermediate feature data, reducing memory accesses and enhancing processing efficiency by sharing a superset of weights across both networks.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If separate neural networks are used for different processing cycles, then each network can be optimized for its specific task, but memory accesses and processing requirements increase
Solution Approach 1:
The patent merges weights from multiple neural networks into a single merged layer that serves multiple processing cycles. This consolidation reduces redundant memory accesses and computational overhead while maintaining the functional capabilities of individual networks through selective weight application.
Solution Approach 2:
The merged layer is designed to serve multiple neural networks and processing cycles universally. By creating a single layer that can be configured with different weight sets, the system achieves multi-functionality without requiring separate dedicated layers for each network, thereby improving computational efficiency.
2Productivity
If a merged layer with combined weights is used, then memory accesses are reduced and processing efficiency improves, but the complexity of weight management increases
Solution Approach 1:
The patent segments the superset of weights into distinct subsets, each corresponding to a specific neural network or processing cycle. This segmentation allows the system to manage complexity by organizing weights in a structured manner while maintaining the benefits of the merged layer architecture.
Solution Approach 2:
The system dynamically selects and applies appropriate weight subsets from the superset based on the current processing cycle requirements. This dynamic weight selection mechanism allows flexible adaptation to different tasks while using a single merged layer, balancing efficiency with manageability.
Data Source
AI summary
A neural network system executable on a processor. The neural network system, when executed on the processor, comprises a merged layer shareable between a first neural network and a second neural network. The merged layer is configured to receive input data from a prior layer of at least one of the first and second neural networks. The merged layer is configured to apply a superset of weights to the input data to generate intermediate feature data representative of at least one feature of the input data, the superset of weights being combined from a first set of weights associated with the first neural network and a second set of weights associated with the second neural network. The merged layer is also configured to output the intermediate feature data to at least one subsequent layer, the at least one subsequent layer serving the first and second neural networks.