Prevalent pre-training strategies for Brain-Computer Interfaces (BCIs) are often constrained by spatio-temporal entanglement. This critical issue arises from processing multi-channel Electroencephalography (EEG) signals as monolithic sequences, which intertwines the signal's temporal dynamics with its spatial topography and hinders the learning of robust and generalizable representations. To address this, we introduce BraSTORM, a framework that explicitly disentangles EEG data into separate temporal and spatial streams at the input level. Two streams are processed by parallel encoders trained with a composite dual-objective: a masked signal reconstruction loss captures fine-grained, intra-modal details, while a cross-modal contrastive loss enforces high-level semantic alignment. Extensive fine-tuning experiments on six benchmarks covering three major BCI downstream tasks—Emotion Recognition, Sleep Staging, and Motor Imagery—demonstrate that BraSTORM achieves state-of-the-art performance. Our findings validate that resolving spatio-temporal entanglement at the input level can be a competitive pre-training framework for the BCI field.

Accurate time series forecasting is crucial across various domains, yet modeling complex, non-stationary time series remains challenging. Recent advancements in diffusion models have inspired their application to time series. However, existing approaches struggle to capture historical patterns and temporal dependencies essential to realize data distributions. Additionally, the reliance on RNN or Transformer architectures intensifies computational costs due to the iterative nature of the diffusion process. To address these shortcomings, we propose a novel diffusion-based model employing a conditional non-autoregressive diffusion process. Specifically, we propose a Historical-Free Guidance (HFG) approach to regulate the denoising process. This approach leverages a state-space-based encoder capable of efficiently extracting latent patterns from historical observations with linear computational complexity. A complementary Temporal encoder is employed to capture temporal dynamics. These representations are then integrated as conditional information and transmitted to a guidance gate that discards the conditional information with a predefined probability. This strategy enables the model to operate seamlessly as both a conditional and unconditional model without added complexity. Comprehensive evaluations conducted on six real-world datasets demonstrate the superior performance of our proposed model compared to established baselines. Notably, the model outperforms the second-best model, achieving average reductions of 11.44% in Mean Squared Error (MSE) and 22.28% in Continuous Ranked Probability Score (CRPS).

Large Language Model (LLM) based multi-agent systems have shown impressive performance across various fields of tasks, further enhanced through collaborative debate and communication using carefully designed communication topologies. However, existing methods typically employ a fixed number of agents or static communication structures, requiring manual pre-definition, and thus struggle to dynamically adapt the number of agents and topology simultaneously to varying task complexities. In this paper, we propose Adaptive Graph Pruning (AGP), a novel task-adaptive multi-agent collaboration framework that jointly optimizes agent quantity (hard-pruning) and communication topology (soft-pruning). Specifically, our method employs a two-stage training strategy: firstly, independently training soft-pruning networks for different agent quantities to determine optimal agent-quantity-specific complete graphs and positional masks across specific tasks; and then jointly optimizing hard-pruning and soft-pruning within a maximum complete graph to dynamically configure the number of agents and their communication topologies per task. Extensive experiments demonstrate that our approach is: (1) High-performing, achieving state-of-the-art results across six benchmarks and consistently generalizes across multiple mainstream LLM architectures, with a increase in performance of 2.58% ∼ 9.84%; (2) Task-adaptive, dynamically constructing optimized communication topologies tailored to specific tasks, with an extremely high performance in all three task categories (general reasoning, mathematical reasoning, and code generation); (3) Token-economical, having fewer training steps and token consumption at the same time, with a decrease in token consumption of 90%+; and (4) Training-efficient, achieving high performance with very few training steps compared with other methods. The performance will surpass the existing baselines after about ten steps of training under six benchmarks. Our code and demos are publicly available at https://resurgamm.github.io/AGP/.

Driver distraction remains a pressing traffic safety issue. Drivers are often careless with their distraction behaviours, which may cause serious traffic accidents. However, current Driver Monitoring Systems (DMS) cannot be put into practical application well, which tend to have high latency, lack precision, and are unable to cover all distraction behaviours. In this paper, we assume driver distraction to be a One-Class Classification (OCC) problem and build an unsupervised learning baseline based on denoising diffusion probabilistic models (DDPM) called RAPID which aggregates future patterns generated by the diffusion process to detect distraction, considering the diversity of normal and abnormal situations. Besides, we propose a skeleton-based synchronized multi-view dataset with diverse distraction behaviours called sktDD (skeleton-based Driver Distraction dataset) to improve on existing datasets. RAPID facilitates a frame-level (0.03 second) and undefined prediction with AUC score beyond State-of-the-Art (SOTA) methods, surpassing currently typical DMS that rely on post-processing procedures and predefined actions. RAPID has the potential to bring significant advancements in the field of traffic safety, which can also be applied in future self-driving scenarios to determine whether the remote-driving operator’s current state is suitable to take over. Our dataset and code are available at https://github.com/jingyulei/rapid.

The transfer of control during the takeover process in conditionally automated driving systems represents a critical safety concern, with the reliability of subsequent driver actions being of paramount importance. Electroencephalography (EEG) data provides a rich source of information regarding driver takeover reliability and is amenable to analysis through deep learning methods. However, current models predominantly focus on the detection of driver physiological and psychological states, neglecting an assessment of driver suitability for takeover from the autonomous system’s perspective. This omission presents a significant missed opportunity for proactive intervention and alert generation. This study introduces a novel approach, the Graph-Based EEG Transformer Network (GETN), which leverages the spatial and temporal characteristics of EEG data. Specifically, GETN employs a Graph Convolutional Network (GCN) as a spatial module to effectively model the topological relationships between distinct brain regions and capture the complex interdependencies between spatially distributed neural sources. Complementing this spatial analysis, a Transformer-based temporal module is integrated to capture temporal dynamics within the EEG signals. The self-attention mechanism elucidates interpretable inter-channel relationships, effectively identifying key channels and associated brain regions instrumental during driver takeover. Utilizing a driver takeover reliability index as the classification label, GETN demonstrates superior performance compared to state-of-the-art EEG signal classification models on a custom-built Takeover Performance (ToP) dataset. Rigorous experimental evaluation demonstrates the superiority of the proposed model, yielding significant improvements in both accuracy and cross-subject generalization.

Ramp metering or navigation guidance plays a vital role in alleviating congestion. However, dynamically integrating ramp metering and navigation guidance to achieve a “1+1>2” effect remains a challenge. To fill the technological gap, this paper proposes a reinforcement learning-based dynamic cooperative ramp metering and navigation guidance (DCRMNG) approach to conduct cooperative traffic control on the expressway system, which simultaneously adjusts demand and supply. Traffic congestion estimation (TCE) and route generation (RG) models are established for future traffic conditions prediction, traffic congestion evaluation, and guiding route generation. To formulate coordinative control strategies for two distinct groups of agents, a heterogeneous-agent Markov decision process (HMDP) is developed. The reward function is carefully crafted to promote collaboration and accelerate the optimization algorithms convergence. Then, a novel heterogeneous-agent proximal policy optimization (HAPPO) algorithm is introduced to solve the DCRMNG approach. Finally, a real-world scenario with an expressway and its parallel arterial road in Hangzhou, China is simulated to assess the performance of the HAPPO-based DCRMNG approach. The results reveal that the proposed approach has the capability to improve the overall network traffic efficiency, mitigate the “navigation jam” issue and achieve intelligent and precise control of the expressway system, exhibiting outstanding performance in various traffic scenarios.

Metro networks necessitate continuous expansion, either extending existing lines or constructing new ones. Optimizing this process, however, presents multifaceted challenges due to complex spatial and demographic relationships, dynamic travel patterns, and a vast solution space with non-linearities and multiple objectives. Existing approaches often fall short, either relying heavily on subjective expert knowledge or limiting their scope to isolated corridors. This paper introduces MetroZero, a deep reinforcement learning (DRL) framework designed to overcome these limitations. We formulate the optimization as a Markov Decision Process (MDP) and leverage a Monte Carlo Tree Search (MCTS) algorithm guided by an actor-critic agent. This powerful combination identifies the optimal sequence of expansion stations within budgetary constraints. To effectively learn network representations, we develop a multiplex graph encoder powered by attentive message passing. A graph attention network (GAT) and a feasibility mask are employed to prioritize high-potential expansion locations and navigate the search space. Inspired by AlphaZero, we train MetroZero through simulated self-play expansion games. Extensive experiments on real-world datasets from Beijing and Changsha demonstrate MetroZero’s effectiveness and superiority. In a complex expansion scenario, it achieves remarkable improvements of 19.6% and 20.4% over the second-best model. Further experiments across varied urban contexts underscore MetorZero’s scalability and adaptability.

Accurately predicting human mobility is crucial for various applications, e.g., transportation services, epidemic control, and advertisement recommendation. Although numerous sequential modeling based methods (e.g., recurrent neural networks) have been proposed for human mobility prediction, accurately modeling individuals’ high-order travel preferences and the influence of social neighbors on their travel decisions remains challenging. In this paper, we construct a novel multi-context aware model for next location prediction, which aggregates multi-dimensional contextual features, including individual preferences, social relations, and activity-location associations. First, we define activity prediction as an auxiliary task and propose an activity-location association pruning method to mitigate the impact of data sparsity on model prediction. Second, we present a novel motif-preserving individual travel preference learning method that leverages a motif-induced hypergraph convolutional network to capture high-order travel preference features explicitly. Third, we identify virtual social neighbors with similar preferences based on individual travel preference learning results, and design a new social gated fusion structure to model the influence of social neighbors on individual travel choices. Finally, experimental results on two real-world travel datasets demonstrate the superiority of the proposed model over baseline models. Our proposed universal method can be seamlessly integrated with other sequential prediction models to improve the accuracy and stability of human mobility prediction.

The urban transportation network is crucial for societal development, but it is prone to failures like congestion caused by accidents or disasters. In particular, often network-wide failure is the result of a series of cascading failures originating from a small set of individual links. To prevent such failures, it is essential to identify these critical links and take early action. However, most existing approaches in the literature for evaluating the importance of each link rely on manually designed metrics (e.g., the Network Robustness Index). These methods are time-consuming and not suitable for large-scale urban networks. Additionally, these metrics fail to accurately capture the dynamic traffic interactions influenced by vehicle movement. In this paper, we present a novel method for identifying critical links by learning effective traffic interaction representation (the spatio-temporal dependencies) among roads. By representing the network as an un-directed graph and abstracting the road links as the nodes, we introduce a temporal graph attention model to capture spatial and temporal dependence between nodes. This model combines a graph attention network and a long short-term memory neural network and produces an attention matrix, which represents traffic interactions among links. Furthermore, we propose a traffic influence propagation model to evaluate the influence of each link for the entire road network based on the traffic interaction representation. We rank the importance of links based on their influence and then identify the critical links. A real-world case study in the city of Hangzhou, China is conducted to test our method and we use the network efficiency ratio to quantify its performance. The results suggest that our method can effectively identify the critical links at different periods.

Recent research has shown that some network traffic flow patterns are similar across multiple traffic regions. Identifying and transferring these domain-invariant features can significantly boost model accuracy and robustness, providing new insights into dealing with modeling issues like traffic data insufficiency and dataset shift. However, how to acquire transferable network traffic flow patterns from multiple traffic regions and adapt such knowledge to downstream prediction tasks of target regions remains challenging. To realize domain-invariant traffic flow pattern transfer and provide more robust prediction under insufficient data conditions, we propose a macroscopic fundamental diagram (MFD) guided transfer learning method, namely physics-guided multi-source domain adversarial network (PG-MDAN). First, an MFD similarity measure is proposed to determine what traffic flow patterns are transferable and to what extent they can be transferred. PG-MDAN embeds this physics-informed transferability measure in domain adversarial pre-training for better adaptation ability. Numerical experiments based on two real-world urban network traffic datasets show that PG-MDAN can successfully transfer recurrent and non-recurrent network traffic flow patterns from multiple regions to provide more robust and responsive prediction performance. Finally, extensive sensitivity analysis is conducted, and the results validate that applying such physical regularization can effectively avoid negative transfer and provide a flexible tool to initiate traffic flow pattern transfer in practice.

Metro bears a large number of passenger flows in urban transportation systems. Short-term metro origin-destination (OD) passenger flow prediction is an essential component of intelligent transportation systems (ITS), which allows operators to better monitor the metro system and improve the level of service for passengers. In this paper, we exploit a novel data structure, hypergraph, to represent the complex correlation between OD pairs, and propose an elaborately designed Spatio-Temporal Dynamic Attentive Multi-HyperGraph Network (ST-DAMHGN) to tackle the short-term OD passenger flow prediction problem. In the proposed framework, we construct multiple hypergraphs to model the relationship between OD pairs and adopt the perceptual field to realize efficient and effective vertex feature extraction. Then, we utilize the attention mechanism to adaptively and dynamically synthesize information from multiple hypergraphs and make a trade-off between exploration and exploitation. A case study is conducted on the metro system of Hangzhou, China. The results of extensive experiments show that ST-DAMHGN outperforms baseline models. The efficiency is validated for the multi-hypergraph model, perceptual field, spatial feature extraction, and attention mechanism. The hypergraph structure used in our model is verified suitable for modeling traffic data without physical road networks to map directly, e.g., OD passenger flow. ST-DAMHGN can be widely implemented by defining proper correlations for model relationships.

Deep Reinforcement Learning has shown some promise in dynamic traffic signal control by adapting to real-time traffic conditions. However, multi-intersection control presents challenges, primarily due to the need for efficient information exchange across increasing intersections, and the importance of spatiotemporal dynamics in traffic flows. Traditional methods often focus solely on spatial or temporal aspects, leading to suboptimal control strategies. This paper introduces a novel Multi-Agent Incentive Communication Deep Reinforcement Learning (MICDRL) method, designed for collaborative control across multiple intersections. MICDRL features an incentive communication mechanism, allowing agents to generate customized messages that influence other agents’ policies, thereby enhancing coordination and achieving globally optimal decisions. A key feature of MICDRL is its reliance on local information for message generation, effectively reducing communication overhead while ensuring collaboration. Additionally, MICDRL integrates a teammate module that leverages temporal data for predicting other agents’ actions, crucial for understanding collective dynamics and spatial environment characteristics. Empirical results show that MICDRL outperforms several state-of-the-art methods in metrics like queue length and throughput. Furthermore, we introduce a tailored three-layer Internet-of-Things architecture to enhance data collection and transmission.

Understanding human drivers’ intentions and predicting their future motions are significant to connected and autonomous vehicles and traffic safety and surveillance systems. Predicting multimodal vehicular trajectories at urban unsignalized intersections remains challenging due to dynamic traffic flow and uncertainty of human drivers’ maneuvers. In this paper, we propose a comprehensive trajectory prediction framework that combines a multimodal trajectory generation network with inverse reinforcement learning (IRL) and risk aversion (RA) modules. Specifically, we first construct a multimodal spatial-temporal Transformer network (mmSTTN) to generate multiple trajectory candidates, using trajectory coordinates as inputs. Accounting for spatio-temporal features, we formulate the IRL reward function for evaluating all candidate trajectories. The optimal trajectory is then selected based on the computed rewards, a process that mimics human drivers’ decision-making. We further develop the RA module based on the driving risk field for optimal risk-averse trajectory prediction. We conduct experiments and ablation studies using the inD dataset at an urban unsignalized intersection, demonstrating impressive human trajectory alignment, prediction accuracy, and the ability to generate risk-averse trajectories. Our proposed framework reduces prediction errors and driving risks by 25% and 30% compared to baseline methods. Results validate vehicles’ human-like risk-averse diverging-and-concentrating behavior as they traverse the intersection. The proposed framework presents a novel approach for forecasting multimodal vehicular trajectories by imitating human drivers and incorporating physics-based risk information derived from the driving field. This research offers a promising direction for enhancing the safety and efficiency of connected and autonomous vehicles navigating urban environments.

Simultaneously and accurately predicting trajectories of multiple heterogeneous agents is crucial for intelligent transportation systems (ITS) applications, e.g., connected and autonomous vehicles. Existing model-based and data-driven methods can achieve good prediction accuracy, but most of them neglect the domain shift issue and prevalent imperfect data problems, i.e., few-shot learning and zero-shot learning issues. To address these issues, we propose a multi-source transfer learning (TL) framework, transferable hierarchical Siamese Transformer network (T-HSTN), for trajectory prediction of multiple heterogeneous agents, e.g., vehicles, bicycles, and pedestrians, at urban unsignalized intersections under small data conditions. Specifically, by extending the self-attention mechanism and exploring feature representations of traffic scenes, a Transformer-based network that hierarchically extracts temporal/spatial features and map features is introduced as the basic prediction model. Moreover, a TL framework with adaptive learning and feature alignment modules is built to explore the feature representations of unfixed traffic scenes and align both statistical and deep features to learn domain-invariant knowledge. More challenging trajectory prediction experiments are designed, corresponding to newly-built or badly-instrumented intersections under real-world scenarios. Experimental results verify the proposed method’s high accuracy, transferability, and generability. Our work fills the gap in solutions and benchmarks for TL tasks in trajectory prediction for heterogeneous agents. The conducted TL experiments provide a more practical setting of considering imperfect data problems in trajectory prediction.

Personal rapid transit (PRT) is a kind of innovative public transport service operated by autonomous vehicles on a dedicated guideway network. In the situation, where PRT guideway network is composed of unidirectional links, the congestion degree of network and vehicle travel distance depend on the arrangement of link directions, given certain traffic demand and network configuration. The motivation of this paper is to propose a method to optimize the link direction of PRT network for enhancing system performance. The proposed optimization problem considers both the vehicle travel time and waiting time at merging nodes as the optimization objective and is formulated as a non-linear programming model. A linear approximation-based solution method is further developed to obtain solutions in an efficient way and the exact solution is achieved via this method. The result shows that the proposed method is effective to find an optimized arrangement of link directions in the network given certain travel demand and network configuration. The computational results of the proposed model in comparison to other models are presented and discussed, which show that the significant reduction of waiting time does not necessarily lead to the high increase of travel time, suggesting the importance of congestion control for merging modes.

This paper presents a novel multiagent approach to automating taxi dispatch that services current bookings in a distributed fashion. The existing system in use by a taxi operator in Singapore and elsewhere, attempts to increase customer satisfaction locally, by sequentially dispatching nearby taxis to service customers. The proposed dispatch system attempts to increase customer satisfaction more globally, by concurrently dispatching multiple taxis to the same number of customers in the same geographical region, andvis-à-vishuman driver satisfaction. To realize the system, a multiagent architecture is proposed, populated with software collaborative agents that can actively negotiate on behalf of taxi drivers in groups of size N for available customer bookings. Theoretically, an analysis of the boundary and optimal multiagent taxi-dispatch situations is presented along with a discussion of their implications. Experimentally, the operational efficiency of the existing and proposed dispatch systems was evaluated through computer simulations. The empirical results, obtained for a 1000-strong taxi fleet over a discrete range of N , show that the proposed system can dispatch taxis with reduction in customer waiting and empty taxi cruising times of up to 33.1% and 26.3%, respectively; and up to 41.8% and 41.2% reduction when a simple negotiation speedup heuristic was applied.

An empirical and comparative evaluation of multiagent taxi dispatch with extended (E) runtime taxi availability is presented. A taxi in operation is said to be E-runtime available if it has a passenger alighting in ?x > 0 minutes' time or is empty, but has no new committed taxi request to service next. In a multiagent architecture, we consider a new operation policy wherein agents of E-runtime available taxis are allowed to negotiate in individual groups of size N for new taxi requests. The main objective is to present an evaluation of the multiagent system performance gains provided by different times-to-arrival of ?x, under a discrete range of demand rates for several iV-group sizes, as compared with the base case when ?x = 0. It is shown that the proposed policy can effectively reduce customer waiting time and empty taxi cruising time by up to about 60% and 96%, respectively, when the service demand is high for a 1000-strong taxi fleet. It is observed that the value selection for the policy parameter ?x is an important aspect for improving the general performance of multiagent taxi dispatch.

This special issue contains revised versions of selected papers originally presented at the 9th IEEE International Conference on Intelligent Transportation Systems (ITSC 2006) held in Toronto, Canada, on September 17-20, 2006.

Transportation planning has proven invaluable in urban planning. As a portion of the cellular traffic is generated by private and public transport passengers during their travels, it is therefore reasonable to argue that transport models can be used as the underlying platform to study the telecommunication traffic generated by this group of high mobility users. This paper looks into this aspect. An introduction on the four-step transport model with feedback is first given. Gravity model based on the maximization of entropy are used to generate the trip distribution. An example is given to illustrate how cell crossing rates can be obtained using the transport model and its usefulness in cellular network planning.