Fuel cell hybrid electric vehicles (FCHEVs) present unique challenges for autonomous driving due to real-time decision-making requirements under multi-objective constraints involving safety, comfort, energy efficiency, and fuel cell durability. Deep reinforcement learning (DRL) offers a promising data-driven framework for such temporally dependent problems, yet its exploratory nature may induce instability and safety risks. To tackle these challenges, this paper proposes an integrated decision optimization and control framework for autonomous FCHEVs. Firstly, a multi-constrained safe reinforcement learning (safe RL) algorithm is developed, incorporating constraints on exploration, experience sampling, and policy updating to achieve collaborative and safe optimization of following distance and energy management through an adaptive multi-objective reward. Additionally, a threshold-based event-triggered backstepping controller is designed to ensure stable tracking of the desired distance and reduce the communication burden. To obtain the optimal backstepping controller with a minimal performance index, an Actor-Critic network structure of DRL is built to approximate the unknown nonlinear term in the Hamilton–Jacobi–Bellman equation, with system stability validated through Lyapunov construction. Simulation results demonstrate that the proposed safe RL-based framework outperforms baseline DRL methods in safety, comfort, and energy efficiency, validating its effectiveness for FCHEVs autonomous driving applications.

Due to the superior ability of global dependency, transformer and its variants have become the primary choice in Masked Time-series Modeling (MTM) towards time-series classification task. In this paper, we experimentally analyze that existing transformer-based MTM methods encounter with two under-explored issues when dealing with time series data: (1) they encode features by performing long-dependency ensemble averaging, which easily results in rank collapse and feature homogenization as the layer goes deeper; (2) they exhibit distinct priorities in fitting different frequency components contained in the time-series, inevitably leading to spectrum energy imbalance of encoded feature. To tackle these issues, we propose an auxiliary content-aware balanced decoder (CBD) to optimize the encoding quality in the spectrum space within masked modeling scheme. Specifically, the CBD iterates on a series of fundamental blocks, and thanks to two tailored units, each block could progressively refine the masked representation via adjusting the interaction pattern based on local content variations of time-series and learning to recalibrate the energy distribution across different frequency components. Moreover, dual-constraint loss is devised to enhance the mutual optimization of vanilla decoder and our CBD. Extensive experimental results on ten time-series classification datasets show that our method nearly surpasses a bunch of baselines. Meanwhile, a series of explanatory results are showcased to sufficiently demystify the behaviors of our method.

Masked Image Modeling (MIM), following “mask-andreconstruct” scheme, is a promising self-supervised method to learn scalable visual representation. Studies indicate that selecting an effective mask strategy is vital for MIM. However, existing approaches often rely on static pre-defined priors, which limit their ability to adapt mask strategies dynamically for network optimization. In this paper, we focus on the learning process of the network and introduce human-like curriculum into MIM for dynamic representation refinement, and propose an end-to-end framework Curriculum Masked Image Modeling (CurMIM). CurMIM consists of two components: Mask Priority Measurer, which acts as a curriculum learner to determine mask priority values using the network’s intrinsic state information, and Dual Adaptive Selector, which serves as a curriculum scheduler to create effective masks based on these values. With negligible extra parameters, our curriculum-based method consistently establishes noticeable improvements across varying model sizes and benchmarks, showing effectiveness and generalization.

Referring Expression Counting (REC) is an emerging task that aims to count specific objects in images based on textual phrases describing their attributes and categories. While current REC baselines inherit architectures from pre-trained open-vocabulary object detectors and demonstrate promising counting and localization capabilities, they overlook critical limitations in the original single-decoder design with shared object queries. This architectural constraint entangles the semantic and localization perception processes, hindering fine-grained understanding of attribute-aware visual features. To address these challenges, we propose DCount, a decoupled counting framework comprising two innovative components: a Decoupled Dual-Decoder (DDD) module and an Attribute Semantic Discriminator (ASD) module. The DDD module separates spatial perception tasks by employing distinct semantic and localization decoders with task-specific object queries, thereby enhancing the capture of discriminative visual features. Building upon the positional and semantic feedback from DDD, the ASD module introduces a two-stage filtering strategy to explicitly mine challenging hard negative attribute samples in the visual domain, while synergistically refining attribute discrimination across both modalities through contrastive learning in the textual domain. Our method achieves state-of-the-art results on both the REC and Zero-Shot Object Counting (ZSOC) benchmarks.

The Internet of Things (IoT) has grown over decades to encompass many forms of sensing modalities, and continues to improve in terms of sophistication and lower costs. The trend of hardware miniaturization and emphasis on user convenience has inspired numerous studies to integrate more varied devices within the IoT into modernizing healthcare systems, facilitating applications, such as activity recognition, fitness assistance, vital signs monitoring, daily dietary tracking, and sleep monitoring. These applications are vital for prevention, detection, and treatment of ailments and can be realized using both dedicated health sensors as well as general-purpose sensors not originally designed for health monitoring. This article surveys such studies, detailing smart health monitoring systems, and the types of sensor components utilized within the IoT. We categorize and analyze these works based on their leverage of device-based techniques (i.e., use of sensors worn or carried by the person) and device-free techniques (i.e., wireless sensing without need to carry hardware), as well as signal processing and classification techniques utilized. In particular, we discuss how different combinations of these techniques can be creatively applied to support professional and commercial health-monitoring IoT networks. We also identify limitations and potential directions that future research may explore.