This repository provides the official implementation of our paper,
Quantum Time-index Models with Reservoir for Time Series Forecasting
To appear at the 31st ACM SIGKDD Conference on Knowledge Discovery and Data Mining (KDD 2025)
We propose Quantum Time-index Models with Reservoir (QuantumTime) — a quantum-classical hybrid model for time series forecasting.
-
🧠 Quantum machine learning module: We integrate variational quantum circuits (VQCs) into the architecture to directly model the high-frequency components of time series. This quantum module acts as an implicit neural representation, enabling compact and expressive feature learning with significantly fewer parameters.
-
🔁 Classical reservoir computing: To capture sequential dependencies and improve extrapolation, we embed a classical reservoir module that provides dynamical memory and rich nonlinearity.
Our model builds upon and extends the framework proposed in the ICML 2023 paper:
Gerald Woo, Chenghao Liu, Doyen Sahoo, Akshat Kumar, and Steven C. H. Hoi.
Learning Deep Time-index Models for Time Series Forecasting.
ICML 2023, Proceedings of Machine Learning Research, Vol. 202.
https://proceedings.mlr.press/v202/woo23b.html
We also draw inspiration from the recent work on quantum implicit neural representations:
Jiaming Zhao, Wenbo Qiao, Peng Zhang, and Hui Gao.
Quantum Implicit Neural Representations.
Forty-first International Conference on Machine Learning (ICML 2024).
https://openreview.net/forum?id=50vc4HBuKU
You can train the model using the following command:
python -m experiments.forecast --config_path=experiments/configs/'dataset_name'/'config_file' run