Trajectory similarity computation is fundamental functionality that is used for, e.g., clustering, prediction, and anomaly detection. However, existing learning-based methods exhibit three key limitations: (1) insufficient modeling of trajectory semantics and hierarchy, lacking both movement dynamics extraction and multi-scale structural representation; (2) high computational costs due to point-wise encoding; and (3) use of physically implausible augmentations that distort trajectory semantics. To address these issues, we propose MovSem, a movement-semantics contrastive learning framework for trajectory similarity computation. MovSem first transforms raw GPS trajectories into movement-semantics features and then segments them into patches. Next, MovSem employs intra- and inter-patch attentions to encode local as well as global trajectory patterns, enabling efficient hierarchical representation and reducing computational costs. Moreover, MovSem includes a curvature-guided augmentation strategy that preserves informative segments (e.g., turns and intersections) and masks redundant ones, generating physically plausible augmented views. Experiments on real-world datasets show that MovSem is capable of outperforming state-of-the-art methods, achieving mean ranks close to the ideal value of 1 at similarity search tasks and improvements by up to 20.3% at heuristic approximation, while reducing inference latency by up to 43.4%.

Unsupervised time series representation learning, driven by recent advances in contrastive learning-based methods, has become a critical component for downstream tasks like forecasting and classification. However, time series data exhibit complex temporal dependencies and spectral patterns, posing challenges for existing approaches to adapt robustly. Moreover, existing contrastive learning-based approaches overlook frequency-domain information and struggle with selecting effective negative samples, further hindering model performance. To address these issues, we propose S a URL-TS, a novel self-adaptive framework for unsupervised time series representation learning. First, it dynamically learns dataset-specific augmentations to generate high-quality positive samples. Second, an adaptive self-supervised learning module with a multi-domain encoder captures both temporal and spectral patterns without relying on negative samples. Third, a representation-wise attention mechanism assigns dynamic weights to representations across domains. To the best of our knowledge, S a URL-TS is the first self-supervised learning framework to jointly model temporal and spectral patterns across both augmentation and learning stages. Extensive experiments confirm the superior performance of S a URL-TS over state-of-the-art models. Notably, its adaptive data augmentation module is plug-and-play and can be integrated into other contrastive learning frameworks, and its learning stage is capable of adapting to a wide range of time series patterns. Our codebase is available at https://github.com/YusenL/SAURL-TS .

Imputation of Correlated Time Series (CTS) is essential in data preprocessing for many tasks, particularly when sensor data is often incomplete. Deep learning has enabled sophisticated models that improve CTS imputation by capturing temporal and spatial patterns. However, deep models often incur considerable consumption of computational resources and thus cannot be deployed in resource-limited settings. This paper presents ReCTSi (Resource-efficient CTS imputation), a method that adopts a new architecture for decoupled pattern learning in two phases: (1) the Persistent Pattern Extraction phase utilizes a multi-view learnable codebook mechanism to identify and archive persistent patterns common across different time series, enabling rapid pattern retrieval during inference. (2) the Transient Pattern Adaptation phase introduces completeness-aware attention modules that allocate attention to the complete and hence more reliable data segments. Extensive experimental results show that ReCTSi achieves state-of-the-art imputation accuracy while consuming much fewer computational resources than the leading existing model, consuming only 0.004% of the FLOPs for inference compared to its closest competitor. The blend of high accuracy and very low resource consumption makes ReCTSi the currently best method for resource-limited scenarios. The related code is available at https://github.com/ryanlaics/RECTSI.

Cyber-physical system sensors emit multivariate time series (MTS) that monitor physical system processes. Such time series generally capture unknown numbers of states, each with a different duration, that correspond to specific conditions, e.g., "walking" or "running" in human-activity monitoring. Unsupervised identification of such states facilitates storage and processing in subsequent data analyses, as well as enhances result interpretability. Existing state-detection proposals face three challenges. First, they introduce substantial computational overhead, rendering them impractical in resourceconstrained or streaming settings. Second, although state-of-the-art (SOTA) proposals employ contrastive learning for representation, insufficient attention to false negatives hampers model convergence and accuracy. Third, SOTA proposals predominantly only emphasize offline non-streaming deployment, we highlight an urgent need to optimize online streaming scenarios. We propose E2Usd that enables efficient-yet-accurate unsupervised MTS state detection. E2Usd exploits a Fast Fourier Transform-based Time Series Compressor (fftCompress) and a Decomposed Dual-view Embedding Module (ddEM) that together encode input MTSs at low computational overhead. Additionally, we propose a False Negative Cancellation Contrastive Learning method (fnccLearning) to counteract the effects of false negatives and to achieve more cluster-friendly embedding spaces. To reduce computational overhead further in streaming settings, we introduce Adaptive Threshold Detection (adaTD). Comprehensive experiments with six baselines and six datasets offer evidence that E2Usd is capable of SOTA accuracy at significantly reduced computational overhead. Our code is available at https://github.com/AI4CTS/E2Usd.

Correlated time series (CTS) forecasting is essential in many practical applications, such as traffic management and server load control. Various deep learning based solutions have been proposed to improve forecasting accuracy. However, while models have become increasingly computationally intensive, they struggle to improve accuracy. This study aims instead to enable more lightweight, accurate models suitable for resource-constrained devices. To achieve this goal, we characterize popular CTS forecasting models, yielding two observations for developing lightweight CTS forecasting. On this basis, we propose theLightCTSframework that adopts plain stacking of temporal and spatial operators instead of alternate stacking which is much more computationally expensive. Moreover,LightCTSfeatures light temporal and spatial operators, L-TCN and GL-Former, offering improved computational efficiency without compromising their feature extraction capabilities.LightCTSalso encompasses a last-shot compression scheme to reduce redundant temporal features and speed up subsequent computations. Next, we equipLightCTSwith two knowledge distillation modules,TafdandCaad, that result inLightCTS$^\star$retaining the original benefits ofLightCTS, while also being able to adapt to varying levels of ultra-constrained resources. Experimental studies offer detailed insight into these proposals and provide evidence that bothLightCTSandLightCTS$^\star$are capable of nearly state-of-the-art accuracy at substantially reduced computational costs.

Correlated time series (CTS) forecasting plays an essential role in many practical applications, such as traffic management and server load control. Many deep learning models have been proposed to improve the accuracy of CTS forecasting. However, while models have become increasingly complex and computationally intensive, they struggle to improve accuracy. Pursuing a different direction, this study aims instead to enable much more efficient, lightweight models that preserve accuracy while being able to be deployed on resource-constrained devices. To achieve this goal, we characterize popular CTS forecasting models and yield two observations that indicate directions for lightweight CTS forecasting. On this basis, we propose the LightCTS framework that adopts plain stacking of temporal and spatial operators instead of alternate stacking that is much more computationally expensive. Moreover, LightCTS features light temporal and spatial operator modules, called L-TCN and GL-Former, that offer improved computational efficiency without compromising their feature extraction capabilities. LightCTS also encompasses a last-shot compression scheme to reduce redundant temporal features and speed up subsequent computations. Experiments with single-step and multi-step forecasting benchmark datasets show that LightCTS is capable of nearly state-of-the-art accuracy at much reduced computational and storage overheads.

Generating a new Chinese font from a multitude of references is an easy task, while it is quite difficult to generate it from a few references. In this paper, we investigate the problem of arbitrary Chinese font generation from a single reference and propose a deep learning based model, named One-reference Chinese Font Generation Network (OCFGNet), to automatically generate any arbitrary Chinese font from a single reference. Based on the disentangled representation learning, we separate the representations of stylized Chinese characters into style and content representations. Then we design a neural network consisting of the style encoder, the content encoder and the joint decoder for the proposed model. The style encoder extracts the style features of style references and maps them onto a continuous Variational Auto-Encoder (VAE) latent variable space while the content encoder extracts the content features of content references and maps them to the content representations. Finally, the joint decoder concatenates both representations in layer-wise to generate the character which has the style of style reference and the content of content reference. In addition, based on Generative Adversarial Network (GAN) structure, we adopt a patch-level discriminator to distinguish whether the received character is real or fake. Besides the adversarial loss, we not only adopt L1-regularized per-pix loss, but also combine a novel loss term Structural SIMilarity (SSIM) together to further drive our model to generate clear and satisfactory results. The experimental results demonstrate that the proposed model can not only extract style and content features well, but also have good performance in the generation of Chinese fonts from a single reference.