Time-Series Foundation Model With Text Embeddings for Sensor Interpretability

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Existing machine learning pipelines face challenges in efficiently processing and analyzing sensory or time-series data, particularly in combining data from different sources and domains, and in providing interpretable results.

Innovation Solution

The development of foundation machine learning models that process time-series data in conjunction with textual descriptions, using techniques such as generating embeddings, combining data with positional embeddings, and employing contextual attention layers to provide controllable attention across sensors and time periods.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Adaptability or versatility

If traditional machine learning models process time-series data from multiple sensors, then the processing complexity increases, but the ability to handle diverse data modalities and provide interpretable results deteriorates

Engineering Contradiction:

Improveability to handle diverse data modalitiesVSAvoidprocessing complexity

Core Design Contradiction:

Adaptability or versatilityVSDevice complexity

Solution Approach 1:

The patent segments the time-series data processing into distinct embedding components: sensor data embeddings, textual description embeddings, and positional embeddings. Each component processes a specific aspect of the input data independently before being combined, which manages complexity while handling diverse modalities effectively

Inventive Principle:

Principle #1Segmentation

Solution Approach 2:

The patent introduces embedding vectors as an intermediary representation that bridges different data modalities (sensor data and textual descriptions). These embeddings serve as a common language that the transformer model can process uniformly, reducing overall system complexity while maintaining versatility

Inventive Principle:

Principle #24Intermediary (Mediator)

Data Source

PatentUS20250200057A1Foundation machine learning models for sensory or other time-series data

Publication Date: 2025.06.19 C3 AI INC

AI summary

A method includes obtaining time-series data of at least one type and at least one textual description of the at least one type of time-series data and processing the time-series data and the at least one textual description using a foundation machine learning model. Processing the time-series data and the at least one textual description using the foundation machine learning model includes generating at least one embedding of the at least one textual description, combining the time-series data and the at least one embedding of the at least one textual description to generate combined data, and generating embedding vectors using the combined data.