Google llc (20240119265). Koopman Neural Forecaster for Time Series with Temporal Distribution Shifts simplified abstract

Koopman Neural Forecaster for Time Series with Temporal Distribution Shifts

Organization Name

google llc

Inventor(s)

Sercan Omer Arik of San Francisco CA (US)

Yihe Dong of New York NY (US)

Qi Yu of San Diego CA (US)

Rui Wang of Mountain View CA (US)

Koopman Neural Forecaster for Time Series with Temporal Distribution Shifts - A simplified explanation of the abstract

This abstract first appeared for US patent application 20240119265 titled 'Koopman Neural Forecaster for Time Series with Temporal Distribution Shifts

Simplified Explanation

The abstract describes a deep sequence model called Koopman Neural Forecaster (KNF) for time series forecasting. KNF utilizes deep neural networks to learn the linear Koopman space and coefficients of chosen measurement functions. It incorporates inductive biases for improved robustness against distributional shifts, utilizing both global and local operators to capture shared characteristics and changing dynamics, respectively. Additionally, KNF includes a feedback loop to continuously update the learned operators over time for rapidly varying behaviors, resulting in superior performance on various time series datasets suffering from distribution shifts.

• KNF is a deep sequence model for time series forecasting.
• It leverages deep neural networks to learn the linear Koopman space and coefficients of chosen measurement functions.
• KNF incorporates inductive biases for improved robustness against distributional shifts.
• It utilizes global and local operators to capture shared characteristics and changing dynamics, respectively.
• KNF includes a feedback loop to continuously update learned operators over time for rapidly varying behaviors.
• KNF demonstrates superior performance on multiple time series datasets suffering from distribution shifts.

Potential Applications

The technology can be applied in various fields such as finance, weather forecasting, energy management, and industrial processes for accurate time series forecasting.

Problems Solved

1. Improved robustness against distributional shifts in time series forecasting. 2. Efficient capturing of changing dynamics in rapidly varying behaviors.

Benefits

1. Superior performance on multiple time series datasets. 2. Continuous updating of learned operators for accurate forecasting. 3. Enhanced robustness against distributional shifts.

Potential Commercial Applications of this Technology

"Deep Sequence Model for Time Series Forecasting: Applications in Finance, Weather Forecasting, and Energy Management"

Possible Prior Art

There may be prior art related to deep sequence models for time series forecasting, such as traditional statistical methods or other deep learning approaches.

How does KNF compare to other deep learning models for time series forecasting?

KNF's performance compared to other deep learning models in terms of accuracy and computational efficiency is not discussed in the abstract.

What are the specific characteristics of the time series datasets used to evaluate KNF's performance?

The abstract does not provide details on the specific time series datasets used to demonstrate KNF's superior performance.

Original Abstract Submitted

aspects of the disclosure provide a deep sequence model, referred to as koopman neural forecaster (knf), for time series forecasting. knf leverages deep neural networks (dnns) to learn the linear koopman space and the coefficients of chosen measurement functions. knf imposes appropriate inductive biases for improved robustness against distributional shifts, employing both a global operator to learn shared characteristics, and a local operator to capture changing dynamics, as well as a specially-designed feedback loop to continuously update the learnt operators over time for rapidly varying behaviors. knf achieves superior performance on multiple time series datasets that are shown to suffer from distribution shifts.