Spiking Neural Network Encoding for Neuromorphic Data Transmission
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing neuromorphic chips face inefficiencies in data processing due to the large amount of pulse data transmission between computing cores, exacerbated by connection and transmission sparsity in spiking neural networks, which consumes significant computing resources and limits energy efficiency.
Innovation Solution
A data processing method and computing core circuit that encodes pulse sequences with consecutive transmitting addresses into route packets, reducing the amount of data to be transmitted and improving processing efficiency by compressing data through encoding and optimizing membrane potential integral operations.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of energy
If pulse data is transmitted directly between computing cores without encoding, then transmission simplicity is maintained, but communication overhead and energy consumption increase significantly
Solution Approach 1:
The patent applies preliminary action by performing encoding processing on pulse sequences before transmission. The encoding module compresses the pulse data in advance, reducing the amount of data that needs to be transmitted between computing cores. This preliminary compression action decreases communication overhead and energy consumption during the actual transmission process.
Solution Approach 2:
The patent changes the parameter representation of pulse data by converting raw pulse sequences into encoded formats. The encoding module transforms the data structure to represent pulse information more efficiently, changing parameters such as data density and transmission format. This parameter transformation reduces the volume of transmitted data while preserving the essential information needed for membrane potential calculations.
2Productivity
If all pulse data is transmitted between computing cores, then complete information is available for processing, but transmission bandwidth and processing time increase
Solution Approach 1:
The patent extracts only the essential information from the pulse sequences for transmission. The encoding module identifies and retains only the critical pulse data needed for membrane potential integral operations, discarding redundant information. This extraction process reduces the amount of data transmitted between computing cores while maintaining the completeness of information necessary for accurate processing.
Solution Approach 2:
The patent segments the pulse data transmission by dividing it into encoded units that can be processed more efficiently. The encoding module breaks down the pulse sequences into manageable segments with consecutive transmitting addresses, allowing for optimized transmission and parallel processing at the receiving end. This segmentation improves productivity by enabling more efficient data handling.
3Quantity of substance
If encoding processing is performed on pulse sequences, then data transmission volume is reduced, but encoding complexity and computational overhead increase
Solution Approach 1:
The patent uses copying by creating encoded representations of the original pulse sequences. The encoding module generates compressed copies of the pulse data that contain the essential information in a more compact form. These encoded copies are then transmitted instead of the original data, reducing transmission volume while the encoding process itself follows systematic patterns that manage complexity.
Solution Approach 2:
The encoding module is designed with multi-functionality to handle various pulse sequence formats and transmission scenarios. By creating a universal encoding mechanism that can process different types of pulse data through standardized operations, the system reduces overall complexity. The same encoding apparatus serves multiple functions: compression, formatting, and preparation for transmission, thereby managing the complexity burden.
Data Source
AI summary
A computing core circuit, including: an encoding module, a route sending module, and a control module, wherein the control module is configured to control the encoding module to perform encoding processing on a pulse sequence determined by pulses of at least one neuron in a current computing core to be transmitted, so as to obtain an encoded pulse sequence, and control the route sending module to determine a corresponding route packet according to the encoded pulse sequence, so as to send the route packet. The present disclosure further provides a data processing method, a chip, a board, an electronic device, and a computer-readable storage medium.