Multi-dimensional Vector Memory Allocation for Digital Banking
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing digital banking systems face challenges in efficiently processing and storing large amounts of multi-dimensional user behavior data, leading to memory inefficiencies and impracticality due to the need for significant memory resources and fixed-length vector requirements in neural networks.
Innovation Solution
A computer-implemented device that enhances memory efficiency by dynamically allocating memory based on user behavior patterns, using multi-dimensional vectors to represent user behavior as points in n-dimensional space, and employing neural networks to process and update these vectors, allowing for efficient data processing and pattern recognition.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If multi-dimensional vectors are used to represent user behavior data, then pattern recognition accuracy is improved, but memory resource requirements increase significantly
Solution Approach 1:
The patent segments the multi-dimensional vector data structure into sparse and dense portions, storing only non-zero elements in a compressed format. This segmentation allows the system to maintain high-dimensional vectors for accurate pattern recognition while significantly reducing memory usage by eliminating storage of zero values.
Solution Approach 2:
The patent introduces a new dimensional organization by representing vectors in n-dimensional space where n exceeds the number of known user attributes. This allows the system to capture complex user behavior patterns across multiple dimensions while using efficient storage schemes that adapt to the sparsity of the data.
2Productivity
If fixed-length vectors are used for neural network processing, then processing efficiency is improved, but memory efficiency deteriorates due to allocation of space for zero or near-zero values
Solution Approach 1:
The patent implements dynamic memory allocation where the storage structure adapts to the actual data sparsity patterns. Instead of static fixed-length allocations, the system dynamically adjusts storage requirements based on the number of non-zero elements, allowing efficient processing while minimizing memory waste.
Solution Approach 2:
The patent changes the storage parameter from fixed-length to variable-length representations. By storing only non-zero elements with their corresponding indices, the system maintains the fixed-length requirement for neural network input while changing the internal storage representation to be more memory-efficient.
3Loss of information
If large amounts of user information are stored to improve insights, then analytical capability is improved, but processing complexity and resource requirements increase
Solution Approach 1:
The patent extracts and stores only the essential non-zero elements from user behavior data, separating meaningful information from redundant zero values. This extraction process reduces processing complexity while maintaining analytical capability by focusing computational resources on actual data patterns rather than processing large amounts of zero padding.
Data Source
AI summary
A computer implemented device is described that is adapted for improving memory efficiency for conducting machine learning on multi-dimensional vectors stored as specially configured data structures. The device includes network interfaces adapted for receiving data sets and communicating with data harvesting applications, as well as data storage and memory for storing the configured data structures. The multi-dimensional vectors and the system provide a data architecture that processes disparate data sets to programmatically extract features transformed from raw information, the extracted features stored in the form of data values suitable for conducting data approximations and neural network processing.