Wenxuan Zhu

Jiangran Lyu, Kai Liu, Xuheng Zhang et al.

Recent robot foundation models largely rely on large-scale behavior cloning, which imitates expert actions but discards transferable dynamics knowledge embedded in heterogeneous embodied data. While the Unified World Model (UWM) formulation has the potential to leverage such diverse data, existing instantiations struggle to scale to foundation-level due to coarse data usage and fragmented datasets. We introduce LDA-1B, a robot foundation model that scales through universal embodied data ingestion by jointly learning dynamics, policy, and visual forecasting, assigning distinct roles to data of varying quality. To support this regime at scale, we assemble and standardize EI-30k, an embodied interaction dataset comprising over 30k hours of human and robot trajectories in a unified format. Scalable dynamics learning over such heterogeneous data is enabled by prediction in a structured DINO latent space, which avoids redundant pixel-space appearance modeling. Complementing this representation, LDA-1B employs a multi-modal diffusion transformer to handle asynchronous vision and action streams, enabling stable training at the 1B-parameter scale. Experiments in simulation and the real world show LDA-1B outperforms prior methods (e.g., $π_{0.5}$) by up to 21\%, 48\%, and 23\% on contact-rich, dexterous, and long-horizon tasks, respectively. Notably, LDA-1B enables data-efficient fine-tuning, gaining 10\% by leveraging 30\% low-quality trajectories typically harmful and discarded.

Jinjie Mai, Wenxuan Zhu, Sara Rojas et al.

Neural radiance fields (NeRFs) generally require many images with accurate poses for accurate novel view synthesis, which does not reflect realistic setups where views can be sparse and poses can be noisy. Previous solutions for learning NeRFs with sparse views and noisy poses only consider local geometry consistency with pairs of views. Closely following \textit{bundle adjustment} in Structure-from-Motion (SfM), we introduce TrackNeRF for more globally consistent geometry reconstruction and more accurate pose optimization. TrackNeRF introduces \textit{feature tracks}, \ie connected pixel trajectories across \textit{all} visible views that correspond to the \textit{same} 3D points. By enforcing reprojection consistency among feature tracks, TrackNeRF encourages holistic 3D consistency explicitly. Through extensive experiments, TrackNeRF sets a new benchmark in noisy and sparse view reconstruction. In particular, TrackNeRF shows significant improvements over the state-of-the-art BARF and SPARF by $\sim8$ and $\sim1$ in terms of PSNR on DTU under various sparse and noisy view setups. The code is available at \href{https://tracknerf.github.io/}.

Wenxuan Zhu, Chao Yu, Qiang Zhang

Deep reinforcement learning (DRL) requires the collection of interventional data, which is sometimes expensive and even unethical in the real world, such as in the autonomous driving and the medical field. Offline reinforcement learning promises to alleviate this issue by exploiting the vast amount of observational data available in the real world. However, observational data may mislead the learning agent to undesirable outcomes if the behavior policy that generates the data depends on unobserved random variables (i.e., confounders). In this paper, we propose two deconfounding methods in DRL to address this problem. The methods first calculate the importance degree of different samples based on the causal inference technique, and then adjust the impact of different samples on the loss function by reweighting or resampling the offline dataset to ensure its unbiasedness. These deconfounding methods can be flexibly combined with existing model-free DRL algorithms such as soft actor-critic and deep Q-learning, provided that a weak condition can be satisfied by the loss functions of these algorithms. We prove the effectiveness of our deconfounding methods and validate them experimentally.

Jianye Yi, Xiaopin Zhong, Weixiang Liu et al.

Semantic segmentation is a classic and fundamental computer vision problem dedicated to assigning each pixel with its corresponding class. Some recent methods introduce edge-based information for improving the segmentation performance. However these methods are specific and limited to certain network architectures, and they can not be transferred to other models or tasks. Therefore, we propose an abstract and universal edge supervision method called Edge-aware Plug-and-play Scheme (EPS), which can be easily and quickly applied to any semantic segmentation models. The core is edge-width/thickness preserving guided for semantic segmentation. The EPS first extracts the Edge Ground Truth (Edge GT) with a predefined edge thickness from the training data; and then for any network architecture, it directly copies the decoder head for the auxiliary task with the Edge GT supervision. To ensure the edge thickness preserving consistantly, we design a new boundarybased loss, called Polar Hausdorff (PH) Loss, for the auxiliary supervision. We verify the effectiveness of our EPS on the Cityscapes dataset using 22 models. The experimental results indicate that the proposed method can be seamlessly integrated into any state-of-the-art (SOTA) models with zero modification, resulting in promising enhancement of the segmentation performance.

Wenxuan Zhu, Bing Li, Cheng Zheng et al.

Multimodal Large Language Models (MLLMs) have demonstrated impressive 2D image/video understanding capabilities. However, there are no publicly standardized benchmarks to assess the abilities of MLLMs in understanding the 4D objects (3D objects with temporal evolution over time). In this paper, we introduce 4D-Bench, the first benchmark to evaluate the capabilities of MLLMs in 4D object understanding, featuring tasks in 4D object Question Answering (4D object QA) and 4D object captioning. 4D-Bench provides 4D objects with diverse categories, high-quality annotations, and tasks necessitating multi-view spatial-temporal understanding, different from existing 2D image/video-based benchmarks. With 4D-Bench, we evaluate a wide range of open-source and closed-source MLLMs. The results from the 4D object captioning experiment indicate that MLLMs generally exhibit weaker temporal understanding compared to their appearance understanding, notably, while open-source models approach closed-source performance in appearance understanding, they show larger performance gaps in temporal understanding. 4D object QA yields surprising findings: even with simple single-object videos, MLLMs perform poorly, with state-of-the-art GPT-4o achieving only 63\% accuracy compared to the human baseline of 91\%. These findings highlight a substantial gap in 4D object understanding and the need for further advancements in MLLMs.

Lu Dai, Wenxuan Zhu, Xuehui Quan et al.

To improve the identification of potential anomaly patterns in complex user behavior, this paper proposes an anomaly detection method based on a deep mixture density network. The method constructs a Gaussian mixture model parameterized by a neural network, enabling conditional probability modeling of user behavior. It effectively captures the multimodal distribution characteristics commonly present in behavioral data. Unlike traditional classifiers that rely on fixed thresholds or a single decision boundary, this approach defines an anomaly scoring function based on probability density using negative log-likelihood. This significantly enhances the model's ability to detect rare and unstructured behaviors. Experiments are conducted on the real-world network user dataset UNSW-NB15. A series of performance comparisons and stability validation experiments are designed. These cover multiple evaluation aspects, including Accuracy, F1- score, AUC, and loss fluctuation. The results show that the proposed method outperforms several advanced neural network architectures in both performance and training stability. This study provides a more expressive and discriminative solution for user behavior modeling and anomaly detection. It strongly promotes the application of deep probabilistic modeling techniques in the fields of network security and intelligent risk control.

Yang Wang, Wenxuan Zhu, Xuehui Quan et al.

This paper addresses the challenges of fault prediction and delayed response in distributed systems by proposing an intelligent prediction method based on temporal feature learning. The method takes multi-dimensional performance metric sequences as input. We use a Gated Recurrent Unit (GRU) to model the evolution of system states over time. An attention mechanism is then applied to enhance key temporal segments, improving the model's ability to identify potential faults. On this basis, a feedforward neural network is designed to perform the final classification, enabling early warning of system failures. To validate the effectiveness of the proposed approach, comparative experiments and ablation analyses were conducted using data from a large-scale real-world cloud system. The experimental results show that the model outperforms various mainstream time-series models in terms of Accuracy, F1-Score, and AUC. This demonstrates strong prediction capability and stability. Furthermore, the loss function curve confirms the convergence and reliability of the training process. It indicates that the proposed method effectively learns system behavior patterns and achieves efficient fault detection.

Nan Jiang, Wenxuan Zhu, Xu Han et al.

This study focuses on the challenge of predicting network traffic within complex topological environments. It introduces a spatiotemporal modeling approach that integrates Graph Convolutional Networks (GCN) with Gated Recurrent Units (GRU). The GCN component captures spatial dependencies among network nodes, while the GRU component models the temporal evolution of traffic data. This combination allows for precise forecasting of future traffic patterns. The effectiveness of the proposed model is validated through comprehensive experiments on the real-world Abilene network traffic dataset. The model is benchmarked against several popular deep learning methods. Furthermore, a set of ablation experiments is conducted to examine the influence of various components on performance, including changes in the number of graph convolution layers, different temporal modeling strategies, and methods for constructing the adjacency matrix. Results indicate that the proposed approach achieves superior performance across multiple metrics, demonstrating robust stability and strong generalization capabilities in complex network traffic forecasting scenarios.

Wenxuan Zhu, Qiyuan Wu, Tengda Tang et al.

This paper addresses the limitations of multi-node perception and delayed scheduling response in distributed systems by proposing a GNN-based multi-node collaborative perception mechanism. The system is modeled as a graph structure. Message-passing and state-update modules are introduced. A multi-layer graph neural network is constructed to enable efficient information aggregation and dynamic state inference among nodes. In addition, a perception representation method is designed by fusing local states with global features. This improves each node's ability to perceive the overall system status. The proposed method is evaluated within a customized experimental framework. A dataset featuring heterogeneous task loads and dynamic communication topologies is used. Performance is measured in terms of task completion rate, average latency, load balancing, and transmission efficiency. Experimental results show that the proposed method outperforms mainstream algorithms under various conditions, including limited bandwidth and dynamic structural changes. It demonstrates superior perception capabilities and cooperative scheduling performance. The model achieves rapid convergence and efficient responses to complex system states.

Shi-Yu Xia, Wenxuan Zhu, Xu Yang et al.

In practice, we usually need to build variable-sized models adapting for diverse resource constraints in different application scenarios, where weight initialization is an important step prior to training. The Learngene framework, introduced recently, firstly learns one compact part termed as learngene from a large well-trained model, after which learngene is expanded to initialize variable-sized models. In this paper, we start from analysing the importance of guidance for the expansion of well-trained learngene layers, inspiring the design of a simple but highly effective Learngene approach termed SWS (Stage-wise Weight Sharing), where both learngene layers and their learning process critically contribute to providing knowledge and guidance for initializing models at varying scales. Specifically, to learn learngene layers, we build an auxiliary model comprising multiple stages where the layer weights in each stage are shared, after which we train it through distillation. Subsequently, we expand these learngene layers containing stage information at their corresponding stage to initialize models of variable depths. Extensive experiments on ImageNet-1K demonstrate that SWS achieves consistent better performance compared to many models trained from scratch, while reducing around 6.6x total training costs. In some cases, SWS performs better only after 1 epoch tuning. When initializing variable-sized models adapting for different resource constraints, SWS achieves better results while reducing around 20x parameters stored to initialize these models and around 10x pre-training costs, in contrast to the pre-training and fine-tuning approach.

Jinjie Mai, Wenxuan Zhu, Haozhe Liu et al.

Reconstructing dynamic 3D scenes (i.e., 4D geometry) from monocular video is an important yet challenging problem. Conventional multiview geometry-based approaches often struggle with dynamic motion, whereas recent learning-based methods either require specialized 4D representation or sophisticated optimization. In this paper, we present Sora3R, a novel framework that taps into the rich spatiotemporal priors of large-scale video diffusion models to directly infer 4D pointmaps from casual videos. Sora3R follows a two-stage pipeline: (1) we adapt a pointmap VAE from a pretrained video VAE, ensuring compatibility between the geometry and video latent spaces; (2) we finetune a diffusion backbone in combined video and pointmap latent space to generate coherent 4D pointmaps for every frame. Sora3R operates in a fully feedforward manner, requiring no external modules (e.g., depth, optical flow, or segmentation) or iterative global alignment. Extensive experiments demonstrate that Sora3R reliably recovers both camera poses and detailed scene geometry, achieving performance on par with state-of-the-art methods for dynamic 4D reconstruction across diverse scenarios.

Bing Li, Cheng Zheng, Wenxuan Zhu et al.

While diffusion models have shown impressive performance in 2D image/video generation, diffusion-based Text-to-Multi-view-Video (T2MVid) generation remains underexplored. The new challenges posed by T2MVid generation lie in the lack of massive captioned multi-view videos and the complexity of modeling such multi-dimensional distribution. To this end, we propose a novel diffusion-based pipeline that generates high-quality multi-view videos centered around a dynamic 3D object from text. Specifically, we factor the T2MVid problem into viewpoint-space and time components. Such factorization allows us to combine and reuse layers of advanced pre-trained multi-view image and 2D video diffusion models to ensure multi-view consistency as well as temporal coherence for the generated multi-view videos, largely reducing the training cost. We further introduce alignment modules to align the latent spaces of layers from the pre-trained multi-view and the 2D video diffusion models, addressing the reused layers' incompatibility that arises from the domain gap between 2D and multi-view data. In support of this and future research, we further contribute a captioned multi-view video dataset. Experimental results demonstrate that our method generates high-quality multi-view videos, exhibiting vivid motions, temporal coherence, and multi-view consistency, given a variety of text prompts.

Chao Yu, Tianpei Yang, Wenxuan Zhu et al.

Providing reinforcement learning agents with informationally rich human knowledge can dramatically improve various aspects of learning. Prior work has developed different kinds of shaping methods that enable agents to learn efficiently in complex environments. All these methods, however, tailor human guidance to agents in specialized shaping procedures, thus embodying various characteristics and advantages in different domains. In this paper, we investigate the interplay between different shaping methods for more robust learning performance. We propose an adaptive shaping algorithm which is capable of learning the most suitable shaping method in an on-line manner. Results in two classic domains verify its effectiveness from both simulated and real human studies, shedding some light on the role and impact of human factors in human-robot collaborative learning.