Neural Network Database Framework for Query Processing
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current database query methods face challenges in efficiently answering queries, particularly nearest neighbor and range aggregate queries, due to the curse of dimensionality and unnecessary calculations, leading to performance degradation and high storage requirements.
Innovation Solution
A neural network database system that uses multiple neural networks trained on partitions of queries to rapidly respond to queries, with a decision tree determining the appropriate neural network for each query, allowing for efficient query processing and reduced storage needs.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If combinatorial algorithms are used to answer database queries, then query processing can be performed, but performance degrades due to the curse of dimensionality and unnecessary calculations
Solution Approach 1:
The patent segments the database into multiple partitions and trains separate neural networks for each partition. This segmentation allows the system to focus computations on relevant data subsets, eliminating unnecessary calculations across the entire database and improving query response time while reducing dimensional curse effects.
Solution Approach 2:
The patent performs preliminary training of neural networks on partitioned database data before actual query processing. This preliminary action pre-computes optimal decision boundaries and patterns, enabling fast query responses without performing unnecessary calculations during actual query execution.
2Measurement precision
If more data is stored to improve query accuracy, then query precision improves, but storage requirements increase
Solution Approach 1:
The patent creates compressed representations of data partitions in the form of trained neural network models. These models capture essential patterns and relationships from the original data, enabling accurate query answers with significantly reduced storage requirements compared to storing all raw data.
Solution Approach 2:
The patent transforms the database from storing raw data points to storing neural network parameters (weights and biases) that represent the data distribution. This parameter transformation maintains query accuracy while dramatically reducing the quantity of stored information.
Data Source
AI summary
A neural network database is disclosed. A learning task to teach a single model to answer any query is formulated. The example neural network database learns existing patterns between query input and output and by exploits the query and data distributions through a decision tree having multiple neural network leaf nodes representing partitions of the queries from the database. The neural network architecture is used to answer different query types efficiently. A generic neural network database framework can learn to answer different query types such as distance to nearest neighbor queries and range aggregate queries. The example neural database answers these two query types with orders of magnitude improvement in query time over the state-of-the-art competitions, and by constructing a model that takes only a fraction of data size.