Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Zero Shot Time Series Forecasting Using Kolmogorov Arnold Networks

Abhiroop Bhattacharya, Nandinee Haq

TL;DR

This work tackles zero-shot electricity price forecasting across unseen markets by learning domain-invariant representations via adversarial cross-domain learning. It fuses Kolmogorov-Arnold Networks (KANs) with a doubly residual N-BEATS-like backbone to decompose time series into backcast and forecast components using spline-based activations, while employing a gradient reversal layer to suppress domain-specific cues. The approach achieves notable improvements over baselines in zero-shot forecasts for Nord Pool and offers interpretable, efficient forecasts suitable for diverse energy markets. The framework holds practical value for robust cross-market forecasting and can be extended with external factors like weather and multiple secondary markets to further enhance generalization.

Abstract

Accurate energy price forecasting is crucial for participants in day-ahead energy markets, as it significantly influences their decision-making processes. While machine learning-based approaches have shown promise in enhancing these forecasts, they often remain confined to the specific markets on which they are trained, thereby limiting their adaptability to new or unseen markets. In this paper, we introduce a cross-domain adaptation model designed to forecast energy prices by learning market-invariant representations across different markets during the training phase. We propose a doubly residual N-BEATS network with Kolmogorov Arnold networks at its core for time series forecasting. These networks, grounded in the Kolmogorov-Arnold representation theorem, offer a powerful way to approximate multivariate continuous functions. The cross domain adaptation model was generated with an adversarial framework. The model's effectiveness was tested in predicting day-ahead electricity prices in a zero shot fashion. In comparison with baseline models, our proposed framework shows promising results. By leveraging the Kolmogorov-Arnold networks, our model can potentially enhance its ability to capture complex patterns in energy price data, thus improving forecast accuracy across diverse market conditions. This addition not only enriches the model's representational capacity but also contributes to a more robust and flexible forecasting tool adaptable to various energy markets.

Zero Shot Time Series Forecasting Using Kolmogorov Arnold Networks

TL;DR

This work tackles zero-shot electricity price forecasting across unseen markets by learning domain-invariant representations via adversarial cross-domain learning. It fuses Kolmogorov-Arnold Networks (KANs) with a doubly residual N-BEATS-like backbone to decompose time series into backcast and forecast components using spline-based activations, while employing a gradient reversal layer to suppress domain-specific cues. The approach achieves notable improvements over baselines in zero-shot forecasts for Nord Pool and offers interpretable, efficient forecasts suitable for diverse energy markets. The framework holds practical value for robust cross-market forecasting and can be extended with external factors like weather and multiple secondary markets to further enhance generalization.

Abstract

Accurate energy price forecasting is crucial for participants in day-ahead energy markets, as it significantly influences their decision-making processes. While machine learning-based approaches have shown promise in enhancing these forecasts, they often remain confined to the specific markets on which they are trained, thereby limiting their adaptability to new or unseen markets. In this paper, we introduce a cross-domain adaptation model designed to forecast energy prices by learning market-invariant representations across different markets during the training phase. We propose a doubly residual N-BEATS network with Kolmogorov Arnold networks at its core for time series forecasting. These networks, grounded in the Kolmogorov-Arnold representation theorem, offer a powerful way to approximate multivariate continuous functions. The cross domain adaptation model was generated with an adversarial framework. The model's effectiveness was tested in predicting day-ahead electricity prices in a zero shot fashion. In comparison with baseline models, our proposed framework shows promising results. By leveraging the Kolmogorov-Arnold networks, our model can potentially enhance its ability to capture complex patterns in energy price data, thus improving forecast accuracy across diverse market conditions. This addition not only enriches the model's representational capacity but also contributes to a more robust and flexible forecasting tool adaptable to various energy markets.

Paper Structure

This paper contains 11 sections, 10 equations, 4 figures, 1 table.

Table of Contents

  1. Introduction
  2. Model Architecture
  3. Dataset
  4. Results
  5. Conclusion
  6. Proposed Framework
  7. Domain Generalization
  8. Proposed Model Backbone
  9. Zero Shot Learning
  10. Covariate Shift
  11. Evaluation Metrics

Figures (4)

  • Figure 1: Line Schematic showing the model architecture consisting of KAN layers stacked together using residual connections inspired by the N-BEATS architecture oreshkin2019n
  • Figure 2: The next day forecast presents a comparison between the KAN, N-BEATS and Proposed model for the NP Market. As indicated, the N-BEATS model produces a smooth forecast while the proposed model uses the flexibility of the B-Spline along with the power of N-BEATS model to produce the best forecast.
  • Figure 3: This representative example shows some of the functions learned by the KAN network when we do zero shot forecasting on the Nord Pool market, using France and Belgium as the primary and secondary markets repsectively.
  • Figure 4: Schematic of our Domain Adaptation Framework.