Awesome-Large-Models-for-Time-Series

SOTA

Venue	Title
	Time-MoE: Billion-Scale Time Series Foundation Models with Mixture of Experts
NeurIPS'24	SOFTS: Efficient Multivariate Time Series Forecasting with Series-Core Fusion
	Are Self-Attentions Effective for Time Series Forecasting?
	In-context Time Series Predictor

Survey

Venue	Title
ICML'24	Position: What Can Large Language Models Tell Us about Time Series Analysis
KDD'24	Foundation Models for Time Series Analysis: A Tutorial and Survey

Training

Pre-Training

Venue	Title	Keywords
ICML'24	Unified Training of Universal Time Series Forecasting Transformers	Multivariate; Large-scale data; Variable window size
ICML’24	Timer: Generative Pre-trained Transformers Are Large Time Series Models	Channel independence; Large-scale data; Auto-regression
arXiv’24	UNITS: A Unified Multi-Task Time Series Model	One model for many tasks; Prompt tuning
ICML'24	MOMENT: A Family of Open Time-series Foundation Models	Masked auto-encoder pretraining
ICML'24	A decoder-only foundation model for time-series forecasting	Auto-regressive patch-wise decoding
ICML'24	TimeSiam: A Pre-Training Framework for Siamese Time-Series Modeling	Reconstruction

Fine-Tuning

Pre-Training & Fine-Tuning

Venue	Title	Keywords
ICML'24	UP2ME: Univariate Pre-training to Multivariate Fine-tuning as a General-purpose Framework for Multivariate Time Series Analysis	Masked auto-encoder pretraining; Variable window size; Multivariate fine-tuning
ICML'24	Multi-Patch Prediction: Adapting LLMs for Time Series Representation Learning	Autoregressive patch-wise decoding

Prompting / Conditional Generation

Venue	Title	Keywords
ICML'24	Time Weaver: A Conditional Time Series Generation Model	Diffusion
ICML'24	S$^2$IP-LLM: Semantic Space Informed Prompt Learning with LLM for Time Series Forecasting	Retrieve word embeddings
KDD'24	POND: Multi-Source Time Series Domain Adaptation with Information-Aware Prompt Tuning	Transfer

Misc

Lightweight

Venue	Title	Keywords
ICML'24	An Analysis of Linear Time Series Forecasting Models	Mathematical proof
ICML'24	SparseTSF: Modeling Long-term Time Series Forecasting with 1k Parameters

Non-stationary

Venue	Title	Keywords
ICML'24	SIN: Selective and Interpretable Normalization for Long-Term Time Series Forecasting

Generalization

Venue	Title	Keywords
ICML'24	Time-Series Forecasting for Out-of-Distribution Generalization Using Invariant Learning

Channel Dependence

Venue	Title	Keywords
ICML'24	CATS: Enhancing Multivariate Time Series Forecasting by Constructing Auxiliary Time Series as Exogenous Variables

Multi-Modal

Venue	Title	Keywords
	MoAT: Multi-Modal Augmented Time Series Forecasting	News article