Lei Liu

Accurate and efficient modeling of agent interactions is essential for trajectory generation in autonomous driving. Existing methods, however, present an “impossible triangle,” failing to simultaneously optimize accuracy, computational time, and memory efficiency. To break this limitation, we first pro pose adapting Rotary Position Embedding (RoPE) from natural language processing to improve relative position encoding in the autonomous driving domain. RoPE's design effectively resolves the time and space complexity overhead of prior query-centric methods. However, we identify that RoPE, while effective for spatial positions, inherently fails to model agent headings due to the periodicity of angular information. To solve this, we propose Directional Rotary Position Embedding (DRoPE), a novel and theoretically-grounded adaptation designed specifically for angles. DRoPE introduces a uniform identity scalar into the 2D rotary transformation, creating a principled mapping that correctly and naturally encodes relative angular information. Our final framework employs a hybrid strategy: using the original RoPE for spatial positions and our proposed DRoPE for agent headings. This combined approach is the first query-centric method to achieve optimal efficiency. Empirical evaluations against state-of-the-art models confirm our framework's practical effectiveness, achieving comparable performance while offering significant computational and memory advantages. The video documentation is available at https://drope-traj.github.io/.

Accurate wind power forecasting is essential for the secure operation of power grids and the integration of large-scale renewable energy. Existing methods primarily exploit dominant meteorological variables with direct physical relationships to power output. However, under highly coupled meteorological conditions, these dominant variables often bias the learned representations toward directly relevant information. This leads to the insufficient exploitation of information from other meteorological variables, thereby limiting further improvements in forecasting performance. To tackle this problem, we propose a dynamics-aware wind power forecasting framework based on latent channel decomposition (LCD), termed Wind-LCD. Specifically, Wind-LCD utilizes learnable latent queries to derive channel-wise representations from the input variables. It then introduces a cosine-similarity regularizer to encourage these representations to be pairwise orthogonal, thereby reducing redundancy and overlap across channels. This design promotes the learning of diverse representations and helps preserve useful information from all meteorological variables. Based on the decomposed representations, the framework employs a dual-branch architecture to model historical observations and future meteorological forecasts, enabling historical temporal dynamics and horizon-specific guidance to be captured before fusion for multi-horizon prediction. Wind-LCD was evaluated on four geographically diverse datasets for 24- to 96-hour forecasting. Results demonstrate that our framework surpasses the existing baselines, achieving average relative reductions of 14.8% in normalized mean absolute error (NMAE) and 12.2% in normalized root mean square error (NRMSE) relative to the best baseline result in each dataset-horizon setting. The source code and datasets will be made publicly available upon publication.

Accurate prediction of tropical cyclone (TC) intensity is essential for effective disaster mitigation. Existing methods mainly rely on limited spatiotemporal information from ERA5 and overlook the spatiotemporal dependencies between historical intensity sequences and spatial environmental features, which restricts their ability to capture the full spatial and temporal patterns needed for intensity forecasting. To address these limitations, we propose the Multi-source Multi-Scale Constrained AutoRegressive (MSCAR) model, the first approach to combine spatiotemporal dependency modeling with large-scale multi-source datasets for autoregressive prediction of global TC intensity. In addition, to fill the gap of a comprehensive dataset covering diverse spatial variables, we introduce the Multi-Source Satellite and ERA5-based Tropical Cyclone Dataset (MSETCD), currently the longest and most complete global dataset for TCs. Experiments show that MSCAR outperforms existing methods, reducing forecast errors globally and regionally by up to 9.52% and 6.74%, respectively. The dataset and source code are openly available at https://github.com/USTC-AI4EEE/MSCAR

Multi-View Clustering (MVC) aggregates heterogeneous data to create a fused representation, which allows it to divide data samples into their respective groups. In recent years, it has drawn increasing attention. However, the current methods have the problem of untrusted fusion. The reasons for this issue are as follows: 1) The existing approaches ignore the fact that the view contains noise data; 2) In deep MVC, the pattern of Contrastive Learning (CL) originates from the same sample as opposed to the identical group. It guides the data fusion process toward an incorrect path. We present a novel Trusted Mixture-of-Mambas Contrastive Network (MoMVC) to address this untrusted problem. The framework is composed of two principal components. Specifically, we develop a new Trusted Mixture-of-Mambas Fusion Network (MoM). It uses the selective mechanism of the Mamba network and the Mixture of Experts (MoE) network to implement the trusted integration of multi-source heterogeneous features. Additionally, we employ the Average-similarity Contrastive Learning (AsCL) module to coordinate the unified embedding with individual view embeddings. By encouraging similarity among view representations at the cluster level rather than solely at the sample level, the AsCL component fundamentally strengthens the reliability of multi-view fusion. Extensive experiments prove that MoMVC establishes new state-of-the-art performance on deep MVC benchmarks, surpassing existing methods by a considerable margin.

Generating trajectories from high-level commands is critical for autonomous driving, but prevailing methods suffer from a flaw we term semantic misalignment. By associating long trajectories with a single, static meta-action (e.g., “lane change”), these methods corrupt training data during maneuver transitions, hindering the learning of robust command-following. To resolve this, we propose the Autoregressive Meta-Action (AMA) framework, a novel formulation that operates at a frame-wise level. It autoregressively predicts the joint distribution of the next-step meta-action and its corresponding state, ensuring strict semantic alignment at every timestep. We implement this framework using a flexible modular architecture where lightweight modules are fine-tuned atop a pre-trained foundation model. This modular design drastically reduces training costs for new commands and enables seamless switching between autonomous and command-following modes. We support our research by releasing an expanded Waymo Open Motion Dataset with dense, frame-level meta-action labels and validate our method’s efficacy on real-world test vehicles.

In recent years, Multi-View Clustering (MVC) has been significantly advanced under the influence of deep learning. By integrating heterogeneous data from multiple views, MVC enhances clustering analysis, making multi-view fusion critical to clustering performance. However, multi-view fusion remains challenged by low-quality data, primarily stemming from two reasons: 1) Certain views are contaminated by noisy data. 2) Some views suffer from missing data. This paper proposes a novel Stochastic Generative Diffusion Fusion (SGDF) method to address this problem. SGDF leverages a multiple generative mechanism for the multi-view feature of each sample. It exhibits robustness against low-quality data. Building on SGDF, we further present the Generative Diffusion Contrastive Network (GDCN). Extensive experiments show that GDCN achieves the state-of-the-art results in deep MVC tasks. The source code is publicly available at https://github.com/HackerHyper/GDCN.