Machine Learning Data Preparation via Spatial Grid Encoding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Machine learning models often struggle to accurately process textual data due to conventional embedding techniques failing to capture context, and numerical data requires additional preparation for effective processing.
Innovation Solution
A machine learning controller is configured to prepare data for processing by encoding spatial data into a grid system, embedding textual data using a deep machine learning model, and harmonizing numerical data, thereby optimizing subsequent processing by a traditional machine learning model.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If conventional embedding techniques are used for textual data, then processing speed is improved, but accuracy deteriorates due to failure to capture context
Solution Approach 1:
The patent introduces spatial encoding as an intermediary representation layer between raw textual data and the machine learning model. Instead of directly embedding text, the system first converts text to spatial coordinates, then encodes these coordinates using grid systems and hierarchical spatial indexes. This intermediary spatial representation preserves contextual relationships while enabling efficient processing by traditional machine learning models.
2Measurement precision
If spatial coordinates are directly processed, then location precision is improved, but device complexity increases due to handling of continuous numerical values
Solution Approach 1:
The patent transforms spatial coordinates from continuous numerical values to discrete categorical representations through parameter changes. By mapping continuous coordinates to grid cell identifiers and hierarchical spatial index levels, the system converts the parameter type from continuous to discrete, making the data suitable for traditional machine learning models while preserving location precision through the hierarchical structure.
Solution Approach 2:
The patent segments the continuous spatial space into discrete grid cells and further divides these cells into hierarchical levels. This segmentation transforms the complex continuous coordinate system into manageable discrete units, reducing processing complexity while maintaining precision through the hierarchical organization that preserves spatial relationships.
3Ease of operation
If hierarchical data is processed separately by level, then processing simplicity is improved, but information completeness deteriorates due to loss of relationships between levels
Solution Approach 1:
The patent merges hierarchical data from multiple levels into a single concatenated feature vector. By combining the hierarchical spatial index values from different levels into one unified representation, the system maintains simplicity in processing while preserving the relationships between hierarchical levels, as the concatenated vector contains information from all levels simultaneously.
Data Source
AI summary
Data for processing by a machine learning model may be prepared by encoding a first portion of the data including a spatial data. The spatial data may include a spatial coordinate including one or more values identifying a geographical location. The encoding of the first portion of the data may include mapping, to a cell in a grid system, the spatial coordinate such that the spatial coordinate is represented by an identifier of the cell instead of the one or more values. The data may be further prepared by embedding a second portion of the data including textual data, preparing a third portion of the data including hierarchical data, and/or preparing a fourth portion of the data including numerical data. The machine learning model may be applied to the prepared data in order to train, validate, test, and/or deploy the machine learning model to perform a cognitive task.