Spatial Temporal Memory System for Multistep Prediction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing predictive analytics software requires extensive user configuration and experience, making it time-consuming and complex to implement advanced features for generating predictive models that accurately forecast values or states multiple time steps into the future.
Innovation Solution
A spatial and temporal memory system that stores relationships between states at earlier times and spatial patterns derived from input data at later times, using a hierarchical structure to predict future states or distributions of values by mapping encoded input data through a spatial pooler and sequence processor, generating sparse vectors and establishing temporal relationships.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If extensive user configuration is used to generate predictive models, then prediction accuracy and model suitability are improved, but implementation time and operational complexity increase
Solution Approach 1:
The system performs self-configuration by automatically learning spatial patterns from input data and generating predictive models without requiring extensive manual user configuration. The spatial pooler and sequence processor automatically extract temporal relationships and generate predictions, eliminating the need for users to manually tune model parameters or configure complex settings.
Solution Approach 2:
The system pre-processes input data by encoding it into spatial patterns and pre-establishes temporal relationships through the sequence processor. This preliminary processing of spatial and temporal information enables the system to generate accurate predictions automatically, reducing the need for post-configuration adjustments and manual intervention.
2Reliability
If extensive user configuration is required for advanced predictive features, then model accuracy is improved, but ease of operation deteriorates
Solution Approach 1:
The predictive system operates autonomously by automatically extracting spatial patterns from input data and generating predictive models without requiring user configuration expertise. The spatial pooler and sequence processor handle all necessary processing tasks automatically, making the system easy to operate while maintaining high prediction accuracy.
Solution Approach 2:
The system divides the predictive modeling process into distinct functional modules: the spatial pooler handles spatial pattern extraction, the sequence processor handles temporal relationship analysis, and the prediction generator creates final predictions. This segmentation allows each component to operate independently and automatically, simplifying user interaction while maintaining comprehensive predictive capabilities.
3Adaptability or versatility
If manual configuration and user operations are used to implement predictive models, then model customization is improved, but device complexity increases
Solution Approach 1:
The system automatically adapts to different input data types and generates customized predictive models without requiring manual configuration. The spatial pooler and sequence processor automatically adjust their processing parameters based on the input data characteristics, enabling model customization while reducing configuration complexity.
Solution Approach 2:
The system dynamically adjusts processing parameters based on the input data and predicted outcomes. The spatial pooler modifies spatial pattern extraction parameters and the sequence processor adjusts temporal relationship parameters automatically, enabling flexible model customization without requiring users to manually change configuration settings.
Data Source
AI summary
Embodiments relate to making predictions for values or states to follow multiple time steps after receiving a certain input data in a spatial and temporal memory system. During a training stage, relationships between states of the spatial and temporal memory system at certain times and spatial patterns of the input data detected a plurality of time steps later after the certain time steps are established. Using the established relationships, the spatial and temporal memory system can make predictions multiple time steps into the future based on the input data received at a current time.