Hierarchical Temporal Memory Feedback Connections
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current Hierarchical Temporal Memory (HTM) systems face challenges in accurately predicting and detecting temporal sequences due to limited contextual information, especially in hierarchical structures where lower processing nodes lack feedback from higher levels, leading to less robust and less accurate predictions.
Innovation Solution
The implementation of feedback connections from upper processing nodes to lower nodes in a hierarchical temporal memory system, allowing lower nodes to determine the inclusion of learned temporal sequences in input data by utilizing sequence inputs and feedback inputs, thereby enhancing prediction and detection accuracy through higher-level contextual information.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If feedback connections are added from upper to lower processing nodes, then prediction and detection accuracy improve, but device complexity increases
Solution Approach 1:
The patent implements feedback connections that transmit information from upper processing nodes back to lower processing nodes. This feedback mechanism allows lower nodes to receive contextual information about temporal sequences and patterns detected at higher levels, enabling them to refine their own detections and predictions. The feedback loops create a hierarchical system where information flows both upward for detection and downward for contextualization, thereby improving overall measurement precision without requiring complete redesign of the network architecture.
2Reliability
If higher-level contextual information is provided to lower nodes, then temporal sequence detection robustness improves, but information processing time increases
Solution Approach 1:
The system performs preliminary processing at lower nodes where computational operations are simpler and faster. Lower processing nodes initially detect basic spatial patterns and temporal sequences without requiring complex contextual information. Only after this preliminary detection do higher-level contextual details become necessary for refinement. This staged approach allows the system to quickly eliminate obvious non-matches while reserving more time-consuming contextual analysis for cases that require it, thereby maintaining robustness without excessive processing time for all inputs.
Data Source
AI summary
Embodiments relate to a first processing node that processes an input data having a temporal sequence of spatial patterns by retaining a higher-level context of the temporal sequence. The first processing node performs temporal processing based at least on feedback inputs received from a second processing node. The first processing node determines whether learned temporal sequences are included in the input data based on sequence inputs transmitted within the same level of a hierarchy of processing nodes and the feedback inputs received from an upper level of the hierarchy of processing nodes.