Canonical Spiking Neuron Network for Spatiotemporal Memory
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current neural systems struggle to effectively integrate and extract relevant information from noisy spatiotemporal patterns, as they lack the ability to efficiently store and retrieve spatiotemporal patterns without explicit instructions or prior knowledge of the patterns' structure.
Innovation Solution
The development of canonical spiking neurons with layered neural net relationships and directional synaptic connectivity, which learn to detect and store spatiotemporal patterns using spike-timing dependent plasticity, allowing for pattern recognition and retrieval in real-time without explicit external instructions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If conventional neural systems are used to process spatiotemporal patterns, then basic signal transmission can be achieved, but the system cannot effectively integrate and extract relevant information from noisy spatiotemporal patterns
Solution Approach 1:
The neural system performs self-organization through unsupervised learning mechanisms. Neurons automatically adjust their synaptic weights based on spike-timing dependent plasticity rules, enabling the system to self-tune to the statistical structure of incoming spatiotemporal patterns without external intervention or explicit programming of pattern recognition capabilities.
Solution Approach 2:
The system dynamically changes synaptic conductance parameters based on the temporal correlation between pre-synaptic and post-synaptic spikes. Through spike-timing dependent plasticity, synaptic weights are continuously adjusted to reflect the statistical properties of the input data, allowing the system to adapt to different spatiotemporal patterns and noise conditions.
2Adaptability or versatility
If the system stores and retrieves spatiotemporal patterns without explicit instructions, then pattern recognition capability is improved, but the complexity of the neural network increases
Solution Approach 1:
The network performs self-organization through unsupervised learning. Neurons automatically adjust their connections based on observed patterns, eliminating the need for external training instructions or complex architectural design. The system self-tunes to the statistical structure of input data through spike-timing dependent plasticity.
Solution Approach 2:
The neural network is organized into distinct functional layers (input layer, hidden layers, output layer) with specific roles. The input layer receives raw spatiotemporal data, hidden layers perform intermediate processing and feature extraction, and the output layer produces recognized patterns. This segmentation allows manageable complexity while achieving sophisticated pattern recognition.
3Productivity
If the system processes noisy spatiotemporal data in real-time, then productivity is improved, but the accuracy of pattern detection decreases
Solution Approach 1:
The neural system continuously processes incoming spatiotemporal data streams without interruption. Neurons maintain persistent spike-generating activity as long as relevant patterns are detected, enabling continuous real-time processing. The system never stops analyzing input data, maintaining constant engagement with the data stream to detect patterns immediately upon their occurrence.
Solution Approach 2:
The system incorporates feedback mechanisms where the output of later layers influences earlier processing stages. Through feedback connections, the network can correct detection errors and refine pattern recognition in real-time. The feedback loop allows the system to adjust its interpretation of noisy input based on contextual information from subsequent processing stages.
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
This solution enables the efficient detection, storage, and retrieval of spatiotemporal patterns in noisy data streams, enhancing the system's ability to recognize patterns and complete partial or corrupted versions, while maintaining stability and noise robustness through strength normalization.
Implementation Method 1
The synaptic conductance can change with time as a function of the relative spike times of pre-synaptic and post-synaptic neurons, as per spike-timing dependent plasticity (STDP). The STDP rule increases the conductance of a synapse if its post-synaptic neuron fires after its pre-synaptic neuron fires, and decreases the conductance of a synapse if the order of the two firings is reversed.
Data Source
AI summary
Embodiments of the invention relate to canonical spiking neurons for spatiotemporal associative memory. An aspect of the invention provides a spatiotemporal associative memory including a plurality of electronic neurons having a layered neural net relationship with directional synaptic connectivity. The plurality of electronic neurons configured to detect the presence of a spatiotemporal pattern in a real-time data stream, and extract the spatiotemporal pattern. The plurality of electronic neurons are further configured to, based on learning rules, store the spatiotemporal pattern in the plurality of electronic neurons, and upon being presented with a version of the spatiotemporal pattern, retrieve the stored spatiotemporal pattern.