This repository serves as a comprehensive backlog and summary of time-series research presented at top computer science conferences. It aims to provide researchers and practitioners with an organized overview of state-of-the-art methods, datasets, and publications across various time-series tasks, including forecasting, imputation, anomaly detection, and generation.

How It Works

The project meticulously curates papers from major conferences like NeurIPS, ICML, ICLR, KDD, AAAI, and IJCAI, categorizing them by task and methodology. It includes details on datasets used, models proposed, publication venue, and available code. The repository also provides a glossary of common time-series abbreviations and links to related resources and other time-series paper collections.

Quick Start & Requirements

• Installation: No direct installation is required as this is a curated list of papers and resources.
• Requirements: Access to academic databases or pre-print servers (like arXiv) may be needed to retrieve the full papers. Links to OneDrive and Google Drive are provided for paper access, though VPN access might be required.
• Resources: The repository itself is lightweight, but accessing the linked papers may require significant bandwidth and storage.

Highlighted Details

• Extensive coverage of recent (2023-2024) and foundational time-series research.
• Categorization by specific tasks (e.g., Multivariable Time Series Forecasting, Time Series Imputation, Anomaly Detection).
• Inclusion of model architectures, datasets, and code availability for many entries.
• Links to related repositories and resources for further exploration.

Maintenance & Community

The repository is maintained by a doctoral student at UNSW, Sydney, with contributions welcomed via issues or pull requests. The author expresses interest in collaboration and discussion on time-series research.

Licensing & Compatibility

The repository itself does not host code or papers directly, but rather links to external resources. The licensing of individual papers and code repositories would need to be checked on their respective platforms.

Limitations & Caveats

The primary limitation is that this is a curated list and not a runnable codebase. Access to some papers may be restricted by paywalls or require specific network access. The organization is primarily driven by the author's research focus, and while comprehensive, might not cover every niche area of time-series analysis.