Related papers: An Exploratory Study to Repurpose LLMs to a Unified Architecture for Time Series Classification

An Exploratory Study to Repurpose LLMs to a Unified Architecture for Time Series Classification

URL: http://arxiv.org/abs/2601.09971v1
Date: Thu, 15 Jan 2026 01:22:53 GMT
Title: An Exploratory Study to Repurpose LLMs to a Unified Architecture for Time Series Classification
Authors: Hansen He, Shuheng Li,
Abstract summary: Inception is the only encoder architecture that consistently yields positive performance gains when integrated with an LLM backbone.<n>This study highlights the impact of time series encoder choice in hybrid LLM architectures and points to Inception-based models as a promising direction for future LLM-driven time series learning.
Score: 0.9126158368122342
License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
Abstract: Time series classification (TSC) is a core machine learning problem with broad applications. Recently there has been growing interest in repurposing large language models (LLMs) for TSC, motivated by their strong reasoning and generalization ability. Prior work has primarily focused on alignment strategies that explicitly map time series data into the textual domain; however, the choice of time series encoder architecture remains underexplored. In this work, we conduct an exploratory study of hybrid architectures that combine specialized time series encoders with a frozen LLM backbone. We evaluate a diverse set of encoder families, including Inception, convolutional, residual, transformer-based, and multilayer perceptron architectures, among which the Inception model is the only encoder architecture that consistently yields positive performance gains when integrated with an LLM backbone. Overall, this study highlights the impact of time series encoder choice in hybrid LLM architectures and points to Inception-based models as a promising direction for future LLM-driven time series learning.

Related papers

TimeLens: Rethinking Video Temporal Grounding with Multimodal LLMs [81.78017865436816]
We present TimeLens, a systematic investigation into building MLLMs with strong video temporal grounding ability.<n>We first expose critical quality issues in existing VTG benchmarks and introduce TimeLens-Bench, comprising meticulously re-annotated versions of three popular benchmarks.<n>We also address noisy training data through an automated re-annotation pipeline, yielding TimeLens-100K, a large-scale, high-quality training dataset.
arXiv Detail & Related papers (2025-12-16T18:59:58Z)
Video-LMM Post-Training: A Deep Dive into Video Reasoning with Large Multimodal Models [78.32948112203228]
Video understanding represents the most challenging frontier in computer vision.<n>Recent emergence of Video-Large Multitemporal Models has demonstrated remarkable capabilities in video understanding tasks.<n>Survey aims to provide researchers and practitioners with a unified framework for advancing Video-LMM capabilities.
arXiv Detail & Related papers (2025-10-06T17:10:44Z)
Time to Embed: Unlocking Foundation Models for Time Series with Channel Descriptions [0.0]
Traditional time series models are task-specific and often depend on dataset-specific training and extensive feature engineering.<n>We introduce $textbfCHARM$, a foundation embedding model for multivariate time series that learns shared, transferable, and domain-aware representations.<n>The model is trained using a Joint Embedding Predictive Architecture (JEPA), with novel augmentation schemes and a loss function designed to improve interpretability and training stability.
arXiv Detail & Related papers (2025-05-20T15:58:54Z)
Logo-LLM: Local and Global Modeling with Large Language Models for Time Series Forecasting [6.537801724497769]
Time series forecasting is critical across multiple domains, where time series data exhibits both local patterns and global dependencies.<n>Recent methods that adapt large language models (LLMs) into time series forecasting inherit this limitation by treating LLMs as black-box encoders.<n>We propose Logo-LLM, a novel LLM-based framework that explicitly extracts and models multi-scale temporal features.
arXiv Detail & Related papers (2025-05-16T09:10:49Z)
TSRM: A Lightweight Temporal Feature Encoding Architecture for Time Series Forecasting and Imputation [1.7819099868722776]
We introduce a temporal feature encoding architecture called Time Series Representation Model (TSRM) for time series forecasting and imputation.<n>The architecture is structured around CNN-based representation layers, each dedicated to an independent representation learning task.<n>The architecture is fundamentally based on a configuration that is inspired by a Transformer encoder, with self-attention mechanisms at its core.
arXiv Detail & Related papers (2025-04-26T09:53:20Z)
LLM-PS: Empowering Large Language Models for Time Series Forecasting with Temporal Patterns and Semantics [56.99021951927683]
Time Series Forecasting (TSF) is critical in many real-world domains like financial planning and health monitoring.<n>Existing Large Language Models (LLMs) usually perform suboptimally because they neglect the inherent characteristics of time series data.<n>We propose LLM-PS to empower the LLM for TSF by learning the fundamental textitPatterns and meaningful textitSemantics from time series data.
arXiv Detail & Related papers (2025-03-12T11:45:11Z)
STORM: Token-Efficient Long Video Understanding for Multimodal LLMs [116.4479155699528]
STORM is a novel architecture incorporating a dedicated temporal encoder between the image encoder and the Video-LLMs.<n>We show that STORM achieves state-of-the-art results across various long video understanding benchmarks.
arXiv Detail & Related papers (2025-03-06T06:17:38Z)
Hierarchical Multimodal LLMs with Semantic Space Alignment for Enhanced Time Series Classification [4.5939667818289385]
HiTime is a hierarchical multi-modal model that seamlessly integrates temporal information into large language models. Our findings highlight the potential of integrating temporal features into LLMs, paving the way for advanced time series analysis.
arXiv Detail & Related papers (2024-10-24T12:32:19Z)
Time-LLM: Time Series Forecasting by Reprogramming Large Language Models [110.20279343734548]
Time series forecasting holds significant importance in many real-world dynamic systems. We present Time-LLM, a reprogramming framework to repurpose large language models for time series forecasting. Time-LLM is a powerful time series learner that outperforms state-of-the-art, specialized forecasting models.
arXiv Detail & Related papers (2023-10-03T01:31:25Z)
UMMAFormer: A Universal Multimodal-adaptive Transformer Framework for Temporal Forgery Localization [16.963092523737593]
We propose a novel framework for temporal forgery localization (TFL) that predicts forgery segments with multimodal adaptation. Our approach achieves state-of-the-art performance on benchmark datasets, including Lav-DF, TVIL, and Psynd.
arXiv Detail & Related papers (2023-08-28T08:20:30Z)
Leaping Into Memories: Space-Time Deep Feature Synthesis [93.10032043225362]
We propose LEAPS, an architecture-independent method for synthesizing videos from internal models. We quantitatively and qualitatively evaluate the applicability of LEAPS by inverting a range of architectures convolutional attention-based on Kinetics-400.
arXiv Detail & Related papers (2023-03-17T12:55:22Z)
Multi-Faceted Representation Learning with Hybrid Architecture for Time Series Classification [16.64345034889185]
We propose a hybrid neural architecture, called Self-Attentive Recurrent Convolutional Networks (SARCoN) SARCoN is the synthesis of long short-term memory networks with self-attentive mechanisms and Fully Convolutional Networks. Our work provides a novel angle that deepens the understanding of time series classification, qualifying our proposed model as an ideal choice for real-world applications.
arXiv Detail & Related papers (2020-12-21T16:42:07Z)

This list is automatically generated from the titles and abstracts of the papers in this site.