EEG-based emotion recognition has become a pivotal area in brain-computer interaction (BCI), offering transformative potential for affective computing. While adaptive graph convolutional networks (AGCNs) excel at modeling spatial relationships in multi-channel EEG signals, and state-of-the-art approaches leverage Transformers for temporal sequence modeling, their integration remains suboptimal. Critical limitations persist in generalizability, computational efficiency, and dynamic spatial-temporal representation—particularly in capturing the non-stationary, long-range dependencies inherent in EEG data. To address these challenges, we propose the ASGC-Mamba, a novel Adaptive Spatial-Temporal Graph Neural Network (AST-GNN) that synergizes enhanced graph learning with state-space sequence modeling. Our framework introduces two key innovations: 1) MultiHead-SAGCN, an adpative graph convolution module that learns heterogeneous topological relationships across multiple latent subspaces, coupled with Spatial Sparsity Learning to eliminate redundant connections and enhance interpretability; 2) A State-Space Temporal Model that replaces conventional Transformers with a Mamba-based architecture, enabling linear-time complexity while maintaining global contextual awareness. Extensive experiments demonstrate the superior performance and computational efficiency of ASGC-Mamba in EEG-based emotion recognition, paving the way for more robust and adaptive BCI applications.

Spatial-temporal data, fundamental to many intelligent applications, reveals dependencies indicating causal links between present measurements at specific locations and historical data at the same or other locations. Within this context, adaptive spatial-temporal graph neural networks (ASTGNNs) have emerged as valuable tools for modelling these dependencies, especially through a data-driven approach rather than pre-defined spatial graphs. While this approach offers higher accuracy, it presents increased computational demands. Addressing this challenge, this paper delves into the concept of localisation within ASTGNNs, introducing an innovative perspective that spatial dependencies should be dynamically evolving over time. We introduce DynAGS, a localised ASTGNN framework aimed at maximising efficiency and accuracy in distributed deployment. This framework integrates dynamic localisation, time-evolving spatial graphs, and personalised localisation, all orchestrated around the Dynamic Graph Generator, a light-weighted central module leveraging cross attention. The central module can integrate historical information in a node-independent manner to enhance the feature representation of nodes at the current moment. This improved feature representation is then used to generate a dynamic sparse graph without the need for costly data exchanges, and it supports personalised localisation. Performance assessments across two core ASTGNN architectures and nine real-world datasets from various applications reveal that DynAGS outshines current benchmarks, underscoring that the dynamic modelling of spatial dependencies can drastically improve model expressibility, flexibility, and system efficiency, especially in distributed settings. © 2025 Owner/Author.

In this paper, we present a novel method to significantly enhance the computational efficiency of Adaptive Spatial-Temporal Graph Neural Networks (ASTGNNs) by introducing the concept of the Graph Winning Ticket (GWT), derived from the Lottery Ticket Hypothesis (LTH). By adopting a pre-determined star topology as a GWT prior to training, we balance edge reduction with efficient information propagation, reducing computational demands while maintaining high model performance. Both the time and memory computational complexity of generating adaptive spatial-temporal graphs is significantly reduced from O(N2) to O(N). Our approach streamlines the ASTGNN deployment by eliminating the need for exhaustive training, pruning, and retraining cycles, and demonstrates empirically across various datasets that it is possible to achieve comparable performance to full models with substantially lower computational costs. Specifically, our approach enables training ASTGNNs on the largest scale spatial-temporal dataset using a single A6000 equipped with 48 GB of memory, overcoming the out-of-memory issue encountered during original training and even achieving state-of-the-art performance. Furthermore, we delve into the effectiveness of the GWT from the perspective of spectral graph theory, providing substantial theoretical support. This advancement not only proves the existence of efficient sub-networks within ASTGNNs but also broadens the applicability of the LTH in resource-constrained settings, marking a significant step forward in the field of graph neural networks. Code is available at https://anonymous.4open.science/r/paper-1430.

Traffic prediction poses a formidable challenge due to the dynamic and intricate spatial-temporal dependencies inherent in the task. Adaptive Spatial-Temporal Graph Neural Networks (ASTGNNs) have emerged as a promising solution, endeavoring to discern node-specific spatial patterns and autonomously deduce the spatial graphs among disparate traffic series. Nonetheless, the efficacy of existing ASTGNNs is often compromised as they struggle to apprehend the dynamic patterns of traffic series and deduce the dynamic spatial graphs in labyrinthine road networks, primarily due to their reliance on static node embeddings. In this paper, we introduce the Adaptive Dynamic Graph Convolutional Recurrent Network (ADGCRN), an innovative ASTGNN archetype adept at learning both dynamic spatial dependencies and dynamic spatial patterns from traffic data. Our model is distinguished by two instance-wise dynamic adaptive modules: i) the Instance-wise Dynamic Adaptive Graph Generation module, which integrates instance-wise state information and temporal periodicity into node embeddings; and ii) the Instance-wise Node Adaptive Parameter Learning module, which is capable of learning dynamic, instance-wise, and node-specific patterns for each traffic series. Rigorous experiments conducted on four benchmark datasets demonstrate that ADGCRN significantly outperforms the state-of-the-art ASTGNNs, achieving an average improvement of 3.45%, 3.11%, and 8.78% in terms of Mean Absolute Error (MAE), Root Mean Square Error (RMSE), and Mean Absolute Percentage Error (MAPE), respectively.

Spatial-temporal graph models are prevailing for abstracting and modelling spatial and temporal dependencies. In this work, we ask the following question: whether and to what extent can we localise spatial-temporal graph models? We limit our scope to adaptive spatial-temporal graph neural networks (ASTGNNs), the state-of-the-art model architecture. Our approach to localisation involves sparsifying the spatial graph adjacency matrices. To this end, we propose Adaptive Graph Sparsification (AGS), a graph sparsification algorithm which successfully enables the localisation of ASTGNNs to an extreme extent (fully localisation). We apply AGS to two distinct ASTGNN architectures and nine spatial-temporal datasets. Intriguingly, we observe that spatial graphs in ASTGNNs can be sparsified by over 99.5% without any decline in test accuracy. Furthermore, even when ASTGNNs are fully localised, becoming graph-less and purely temporal, we record no drop in accuracy for the majority of tested datasets, with only minor accuracy deterioration observed in the remaining datasets. However, when the partially or fully localised ASTGNNs are reinitialised and retrained on the same data, there is a considerable and consistent drop in accuracy. Based on these observations, we reckon that (i) in the tested data, the information provided by the spatial dependencies is primarily included in the information provided by the temporal dependencies and, thus, can be essentially ignored for inference; and (ii) although the spatial dependencies provide redundant information, it is vital for the effective training of ASTGNNs and thus cannot be ignored during training. Furthermore, the localisation of ASTGNNs holds the potential to reduce the heavy computation overhead required on large-scale spatial-temporal data and further enable the distributed deployment of ASTGNNs.

Time series forecasting has widespread applications in urban life ranging from air quality monitoring to traffic analysis. However, accurate time series forecasting is challenging because real-world time series suffer from the distribution shift problem, where their statistical properties change over time. Despite extensive solutions to distribution shifts in domain adaptation or generalization, they fail to function effectively in unknown, constantly-changing distribution shifts, which are common in time series. In this paper, we propose Hyper TimeSeries Forecasting (HTSF), a hypernetwork-based framework for accurate time series forecasting under distribution shift. HTSF jointly learns the time-varying distributions and the corresponding forecasting models in an end-to-end fashion. Specifically, HTSF exploits the hyper layers to learn the best characterization of the distribution shifts, generating the model parameters for the main layers to make accurate predictions. We implement HTSF as an extensible framework that can incorporate diverse time series forecasting models such as RNNs. Extensive experiments on 7 benchmarks demonstrate that HTSF achieves state-of-the-art performances.

Pre-trained language models have been widely adopted as backbones in various natural language processing tasks.However, existing pre-trained language models ignore the descriptive meta-information in the text such as the distinction between the title and the mainbody, leading to over-weighted attention to insignificant text.In this paper, we propose a hypernetwork-based architecture to model the descriptive meta-information and integrate it into pre-trained language models.Evaluations on three natural language processing tasks show that our method notably improves the performance of pre-trained language models and achieves the state-of-the-art results on keyphrase extraction.