TimesJev speeds up time series reasoning while preserving accuracy
TimesJev: Thinking Fast on Time Series with System 1 Sufficiency
Artificial Intelligence
Summary
Time series data, like heartbeats or stock prices, can be hard to analyze quickly because they have many numbers over time. The authors created TimesJev, a fast method that compresses important patterns in time series before analyzing them. This lets the model make decisions quickly without losing much important information. They tested TimesJev on many datasets and found it was much faster and more accurate than other models that take longer to think through the data.
What this means in practice
- •For time series analysts: Process large time series datasets faster with compressed representations that retain key information for quicker insights.
- •For real-time data engineers: Deploy low-latency time series models that make instant decisions using key frequency patterns for applications like monitoring or anomaly detection.
Authors
Chenghao Liu, Wei Fan, Jindong Han, Dongjie Wang, Aniket Mahanti, Pengyang Wang, Hao Liu
Abstract
Large language models (LLMs) have recently shown strong potential for time series reasoning, while models that reason over time series are usually with long chains of thought. However, this process often consumes thousands of tokens, directly leading to substantial latency during inference. In contrast, System 1 models such as Jev make decisions instantly and can naturally be an alternative, but raw time series are not compatible with the embedding space of these models, thus hindering the reasoning performance, especially when input series contain thousands of numeric points. In this paper, we make a very early attempt to propose a time series System 1 model, namely TimesJev. Specifically, we first formalize System 1 sufficiency in time series reasoning, which motivates a compressed representation for time series with key frequency patterns preserved. To construct this representation, we derive a dimension reduction objective that combines (i) a sufficiency divergence that measures the core information lost during compression, and (ii) a realizability divergence that quantifies the incompatibility of the System 1 model to the optimal prediction given this representation. We optimize this objective by learning time series compression with a gating mechanism for sufficiency, and then aligning the selected representation with the embedding space for realizability. Finally, experiments on 13 datasets demonstrate that TimesJev outperforms reasoning models by over 62% with 62.9% less inference time than their fine-tuned versions on average.