Constrained decision-making problems are pervasive across various disciplines.

By introducing LLM-embedded textual timestamps, Auto-Times can utilize chronological information to align multivariate time series. Empirically, AutoTimes achieves state-of-the-art with 0.1% trainable parameters and

Lately, there has been a surge in interest surrounding generative modeling of time series data. Most existing approaches are designed either to process short sequences or to handle long-range sequences. This dichotomy can be attributed to gradient issues with recurrent networks, computational costs associated with transformers, and limited expressiveness of state space models. Towards a unified generative model for varying-length time series, we propose in this work to transform sequences into images.

Intermittent control problems are common in real world.

Conformal prediction is a powerful tool for uncertainty qua ntification, but its application to time-series data is constrained by the violati on of the exchangeability assumption. Current solutions for time-series prediction typically operate in the output space and rely on manually selected weights to addres s distribution drift, leading to overly conservative predictions. To enable dyna mic weight learning in the semantically rich latent space, we introduce a novel a pproach called Con-formalized Time Series with Semantic Features (CT -SSF). CT -SSF utilizes the inductive bias in deep representation learning to dynamica lly adjust weights, prioritizing semantic features relevant to the current predic tion. Theoretically, we show that CT -SSF surpasses previous methods defined in the ou tput space. Experiments on synthetic and benchmark datasets demonstrate tha t CT -SSF significantly outperforms existing state-of-the-art (SOT A) conformal p rediction techniques in terms of prediction efficiency while maintaining a valid cov erage guarantee.

This highlights the importance of selecting an appropriate schedule for each dataset.

Recently powerful Large Language Models (LLMs) have become easily accessible to hundreds of millions of users world-wide.

Despite the recent strides in crafting specific architectures for time-series forecasting and developing pre-trained universal models, a comprehensive examination of their capability in accommodating varied-horizon forecasting during inference is still lacking.

It solves distribution-valued supervised learning, where the output values of the training dataset are probability distributions.