Temporal Vector Embedding Arrays for Incremental Model Training
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Generative AI models lack temporal awareness in vector embeddings, limiting their ability to perform incremental training and analyze how data sets evolve over time.
Innovation Solution
Integrate temporal data with vector embeddings arrays, allowing for the storage and comparison of vector embeddings data sets over different time periods to enable temporal-aware model training.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If vector embeddings are used as input representation for generative AI models, then the model can process various types of data efficiently, but the model lacks temporal awareness and cannot analyze data evolution over time
Solution Approach 1:
The patent combines vector embeddings with temporal data structures (timestamps, time periods, sequence information) to create a hybrid representation that preserves both the semantic meaning of the data and its temporal context. This merging allows the system to maintain the versatility of vector embeddings while adding temporal awareness through structured temporal metadata.
2Productivity
If snapshots of information are taken as input data sets, then the model can process data efficiently, but incremental training is limited and the model cannot compare deltas between data sets over time
Solution Approach 1:
The patent applies preliminary action by pre-processing data into both vector embedding representations and structured temporal data formats during data ingestion. This preliminary structuring of temporal information (timestamps, time periods, sequence identifiers) enables efficient incremental training later, as the temporal framework is already in place for comparing data sets across different time periods without requiring complex real-time processing.
3Loss of information
If temporal data is added to vector embeddings to enable temporal analysis, then the model can perform incremental training and analyze data evolution, but the data structure becomes more complex
Solution Approach 1:
The patent segments temporal information into distinct, standardized components (timestamps, time periods, sequence identifiers, temporal relationships) that are separately structured and stored alongside vector embeddings. This segmentation allows the system to retain comprehensive temporal information while managing complexity through modular organization, where each temporal element can be independently processed and queried.
Data Source
AI summary
A system may include a storage device. The system may further include a plurality of processing node in communication with the storage device. At least one processing node of the plurality of processing nodes may receive a data set from a data source. The at least one processing node may execute a model on the received data set to generate a vector embeddings array representative of the received data. The at least one processing node may identify temporal data associated with the vector embeddings array. The at least one processing node may store the vector embeddings array with the associated temporal data in the storage device. A method and computer-readable medium are also disclosed.