Serial Neural Network Engine Bus Architecture
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing neural networks require extensive memory resources due to numerous connections between nodes, especially in wide networks with many hidden layers, and often necessitate parallel processing across multiple processors, leading to high computational power requirements and inefficiencies in handling large data sets.
Innovation Solution
A serial neural network engine utilizes a common serial bus to transmit data to nodes, allowing each node to determine whether to store data in its register based on source identification, eliminating the need for external memory and reducing computational power requirements by processing data in a serial manner.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If parallel processing across multiple processors is used to handle large data sets, then processing capability is improved, but memory resources and computational power requirements increase significantly
Solution Approach 1:
The system segments the data processing task by dividing input data into individual elements that are transmitted serially to multiple nodes. Each node processes only the specific data element assigned to it based on source identification, eliminating the need for all nodes to access and store the entire data set simultaneously. This segmentation allows parallel processing capability while reducing overall memory requirements.
Solution Approach 2:
The system performs preliminary action by pre-configuring each node with its designated source identification data before the serial transmission begins. When data elements are transmitted serially, each node can immediately determine whether to store or process the incoming data based on pre-established matching criteria, eliminating the need for complex runtime memory management and reducing memory overhead.
2Adaptability or versatility
If extensive connections between nodes are implemented to handle wide networks with many hidden layers, then network capability is improved, but memory consumption increases
Solution Approach 1:
The system extracts and transmits only the necessary data elements serially through the common bus rather than maintaining extensive parallel connections between all node pairs. Each node extracts and stores only the data elements relevant to its processing function, determined by source identification matching. This extraction approach enables complex network capabilities while minimizing memory consumption by eliminating redundant data storage.
3Quantity of substance
If a common serial bus is used to transmit data serially, then memory consumption is reduced, but data transmission speed may be limited
Solution Approach 1:
The system implements self-service by enabling each node to autonomously determine whether to store incoming serial data based on source identification matching. Nodes independently filter and process only their required data elements without requiring centralized control or additional buffering, allowing efficient serial transmission while minimizing memory usage. The autonomous decision-making at each node compensates for the slower serial transmission rate.
Data Source
AI summary
Embodiments of serial neural network configuration and processing via a common serial bus are disclosed. In some embodiments, the input data and source identification data is sent to nodes of the neural network serially. The nodes can determine whether the source identification data matches with an address for the node. If the address matches, the node can store the input data in its register for further processing. In some embodiments, the serial neural network engine can include a common serial bus that can broadcast data across multiple processor chips or cores.