Wenjing Bian, Zirui Wang, Kejie Li, Victor Adrian Prisacariu
We propose Ray-ONet to reconstruct detailed 3D models from monocular images efficiently. By predicting a series of occupancy probabilities along a ray that is back-projected from a pixel in the camera coordinate, our method Ray-ONet improves the reconstruction accuracy in comparison with Occupancy Networks (ONet), while reducing the network inference complexity to O($N^2$). As a result, Ray-ONet achieves state-of-the-art performance on the ShapeNet benchmark with more than 20$\times$ speed-up at $128^3$ resolution and maintains a similar memory footprint during inference.
Zirui Wang, Adams Wei Yu, Orhan Firat, Yuan Cao
This paper explores zero-label learning in Natural Language Processing (NLP), whereby no human-annotated data is used anywhere during training and models are trained purely on synthetic data. At the core of our framework is a novel approach for better leveraging the powerful pretrained language models. Specifically, inspired by the recent success of few-shot inference on GPT-3, we present a training data creation procedure named Unsupervised Data Generation (UDG), which leverages few-shot prompts to synthesize high-quality training data without real human annotations. Our method enables zero-label learning as we train task-specific models solely on the synthetic data, yet we achieve better or comparable results from strong baseline models trained on human-labeled data. Furthermore, when mixed with labeled data, our approach serves as a highly effective data augmentation procedure, achieving new state-of-the-art results on the SuperGLUE benchmark.
Zirui Wang, Mengzhou Xia, Luxi He, Howard Chen, Yitao Liu, Richard Zhu, Kaiqu Liang, Xindi Wu, Haotian Liu, Sadhika Malladi, Alexis Chevalier, Sanjeev Arora, Danqi Chen
Chart understanding plays a pivotal role when applying Multimodal Large Language Models (MLLMs) to real-world tasks such as analyzing scientific papers or financial reports. However, existing datasets often focus on oversimplified and homogeneous charts with template-based questions, leading to an over-optimistic measure of progress. We demonstrate that although open-source models can appear to outperform strong proprietary models on these benchmarks, a simple stress test with slightly different charts or questions can deteriorate performance by up to 34.5%. In this work, we propose CharXiv, a comprehensive evaluation suite involving 2,323 natural, challenging, and diverse charts from arXiv papers. CharXiv includes two types of questions: 1) descriptive questions about examining basic chart elements and 2) reasoning questions that require synthesizing information across complex visual elements in the chart. To ensure quality, all charts and questions are handpicked, curated, and verified by human experts. Our results reveal a substantial, previously underestimated gap between the reasoning skills of the strongest proprietary model (i.e., GPT-4o), which achieves 47.1% accuracy, and the strongest open-source model (i.e., InternVL Chat V1.5), which achieves 29.2%. All models lag far behind human performance of 80.5%, underscoring weaknesses in the chart understanding capabilities of existing MLLMs. We hope CharXiv facilitates future research on MLLM chart understanding by providing a more realistic and faithful measure of progress. Project page and leaderboard: https://charxiv.github.io/
Zirui Wang, Zhizhou Sha, Zheng Ding, Yilin Wang, Zhuowen Tu
We present TokenCompose, a Latent Diffusion Model for text-to-image generation that achieves enhanced consistency between user-specified text prompts and model-generated images. Despite its tremendous success, the standard denoising process in the Latent Diffusion Model takes text prompts as conditions only, absent explicit constraint for the consistency between the text prompts and the image contents, leading to unsatisfactory results for composing multiple object categories. TokenCompose aims to improve multi-category instance composition by introducing the token-wise consistency terms between the image content and object segmentation maps in the finetuning stage. TokenCompose can be applied directly to the existing training pipeline of text-conditioned diffusion models without extra human labeling information. By finetuning Stable Diffusion, the model exhibits significant improvements in multi-category instance composition and enhanced photorealism for its generated images. Project link: https://mlpc-ucsd.github.io/TokenCompose
Zirui Wang, Zhenxi Song, Yi Guo, Yuxin Liu, Guoyang Xu, Min Zhang, Zhiguo Zhang
The development of EEG decoding algorithms confronts challenges such as data sparsity, subject variability, and the need for precise annotations, all of which are vital for advancing brain-computer interfaces and enhancing the diagnosis of diseases. To address these issues, we propose a novel two-stage approach named Self-Supervised State Reconstruction-Primed Riemannian Dynamics (EEG-ReMinD) , which mitigates reliance on supervised learning and integrates inherent geometric features. This approach efficiently handles EEG data corruptions and reduces the dependency on labels. EEG-ReMinD utilizes self-supervised and geometric learning techniques, along with an attention mechanism, to analyze the temporal dynamics of EEG features within the framework of Riemannian geometry, referred to as Riemannian dynamics. Comparative analyses on both intact and corrupted datasets from two different neurodegenerative disorders underscore the enhanced performance of EEG-ReMinD.
Zirui Wang
We establish global-in-time decay estimates for the multi-phase Muskat problem in the case where the density takes exactly n+1 distinct constant values. We first linearize the system around a flat stable configuration, followed by the study of associated linearized operator. The asymptotic behavior at low frequencies of eigenvalues yields the decay rate of (1+t)^{-s/2-1/4} for Wiener norm \|f\|_s, which is slower than the classical case, where the decay rate is (1+t)^{-s+ν}. Afterwards we bound the nonlinear term to close the argument.
Zirui Wang, Zachary C. Lipton, Yulia Tsvetkov
Modern multilingual models are trained on concatenated text from multiple languages in hopes of conferring benefits to each (positive transfer), with the most pronounced benefits accruing to low-resource languages. However, recent work has shown that this approach can degrade performance on high-resource languages, a phenomenon known as negative interference. In this paper, we present the first systematic study of negative interference. We show that, contrary to previous belief, negative interference also impacts low-resource languages. While parameters are maximally shared to learn language-universal structures, we demonstrate that language-specific parameters do exist in multilingual models and they are a potential cause of negative interference. Motivated by these observations, we also present a meta-learning algorithm that obtains better cross-lingual transferability and alleviates negative interference, by adding language-specific layers as meta-parameters and training them in a manner that explicitly improves shared layers' generalization on all languages. Overall, our results show that negative interference is more common than previously known, suggesting new directions for improving multilingual representations.
Zirui Wang, Jiateng Xie, Ruochen Xu, Yiming Yang, Graham Neubig, Jaime Carbonell
Learning multilingual representations of text has proven a successful method for many cross-lingual transfer learning tasks. There are two main paradigms for learning such representations: (1) alignment, which maps different independently trained monolingual representations into a shared space, and (2) joint training, which directly learns unified multilingual representations using monolingual and cross-lingual objectives jointly. In this paper, we first conduct direct comparisons of representations learned using both of these methods across diverse cross-lingual tasks. Our empirical results reveal a set of pros and cons for both methods, and show that the relative performance of alignment versus joint training is task-dependent. Stemming from this analysis, we propose a simple and novel framework that combines these two previously mutually-exclusive approaches. Extensive experiments demonstrate that our proposed framework alleviates limitations of both approaches, and outperforms existing methods on the MUSE bilingual lexicon induction (BLI) benchmark. We further show that this framework can generalize to contextualized representations such as Multilingual BERT, and produces state-of-the-art results on the CoNLL cross-lingual NER benchmark.
Zirui Wang, Jiahui Yu, Adams Wei Yu, Zihang Dai, Yulia Tsvetkov, Yuan Cao
With recent progress in joint modeling of visual and textual representations, Vision-Language Pretraining (VLP) has achieved impressive performance on many multimodal downstream tasks. However, the requirement for expensive annotations including clean image captions and regional labels limits the scalability of existing approaches, and complicates the pretraining procedure with the introduction of multiple dataset-specific objectives. In this work, we relax these constraints and present a minimalist pretraining framework, named Simple Visual Language Model (SimVLM). Unlike prior work, SimVLM reduces the training complexity by exploiting large-scale weak supervision, and is trained end-to-end with a single prefix language modeling objective. Without utilizing extra data or task-specific customization, the resulting model significantly outperforms previous pretraining methods and achieves new state-of-the-art results on a wide range of discriminative and generative vision-language benchmarks, including VQA (+3.74% vqa-score), NLVR2 (+1.17% accuracy), SNLI-VE (+1.37% accuracy) and image captioning tasks (+10.1% average CIDEr score). Furthermore, we demonstrate that SimVLM acquires strong generalization and transfer ability, enabling zero-shot behavior including open-ended visual question answering and cross-modality transfer.
Zirui Wang, Xinran Zhao, Simon Stepputtis, Woojun Kim, Tongshuang Wu, Katia Sycara, Yaqi Xie
Understanding and predicting human actions has been a long-standing challenge and is a crucial measure of perception in robotics AI. While significant progress has been made in anticipating the future actions of individual agents, prior work has largely overlooked a key aspect of real-world human activity -- interactions. To address this gap in human-like forecasting within multi-agent environments, we present the Hierarchical Memory-Aware Transformer (HiMemFormer), a transformer-based model for online multi-agent action anticipation. HiMemFormer integrates and distributes global memory that captures joint historical information across all agents through a transformer framework, with a hierarchical local memory decoder that interprets agent-specific features based on these global representations using a coarse-to-fine strategy. In contrast to previous approaches, HiMemFormer uniquely hierarchically applies the global context with agent-specific preferences to avoid noisy or redundant information in multi-agent action anticipation. Extensive experiments on various multi-agent scenarios demonstrate the significant performance of HiMemFormer, compared with other state-of-the-art methods.
Zirui Wang, Wenjing Bian, Xinghui Li, Yifu Tao, Jianeng Wang, Maurice Fallon, Victor Adrian Prisacariu
We introduce Oxford Day-and-Night, a large-scale, egocentric dataset for novel view synthesis (NVS) and visual relocalisation under challenging lighting conditions. Existing datasets often lack crucial combinations of features such as ground-truth 3D geometry, wide-ranging lighting variation, and full 6DoF motion. Oxford Day-and-Night addresses these gaps by leveraging Meta ARIA glasses to capture egocentric video and applying multi-session SLAM to estimate camera poses, reconstruct 3D point clouds, and align sequences captured under varying lighting conditions, including both day and night. The dataset spans over 30 $\mathrm{km}$ of recorded trajectories and covers an area of 40,000 $\mathrm{m}^2$, offering a rich foundation for egocentric 3D vision research. It supports two core benchmarks, NVS and relocalisation, providing a unique platform for evaluating models in realistic and diverse environments.
Zirui Wang, Zihang Dai, Barnabás Póczos, Jaime Carbonell
When labeled data is scarce for a specific target task, transfer learning often offers an effective solution by utilizing data from a related source task. However, when transferring knowledge from a less related source, it may inversely hurt the target performance, a phenomenon known as negative transfer. Despite its pervasiveness, negative transfer is usually described in an informal manner, lacking rigorous definition, careful analysis, or systematic treatment. This paper proposes a formal definition of negative transfer and analyzes three important aspects thereof. Stemming from this analysis, a novel technique is proposed to circumvent negative transfer by filtering out unrelated source data. Based on adversarial networks, the technique is highly generic and can be applied to a wide range of transfer learning algorithms. The proposed approach is evaluated on six state-of-the-art deep transfer methods via experiments on four benchmark datasets with varying levels of difficulty. Empirically, the proposed method consistently improves the performance of all baseline methods and largely avoids negative transfer, even when the source data is degenerate.
Zirui Wang
Transfer learning has been proven effective when within-target labeled data is scarce. A lot of works have developed successful algorithms and empirically observed positive transfer effect that improves target generalization error using source knowledge. However, theoretical analysis of transfer learning is more challenging due to the nature of the problem and thus is less studied. In this report, we do a survey of theoretical works in transfer learning and summarize key theoretical guarantees that prove the effectiveness of transfer learning. The theoretical bounds are derived using model complexity and learning algorithm stability. As we should see, these works exhibit a trade-off between tight bounds and restrictive assumptions. Moreover, we also prove a new generalization bound for the multi-source transfer learning problem using the VC-theory, which is more informative than the one proved in previous work.
Chen Yao, Yangtao Ge, Guowei Shi, Zirui Wang, Ningbo Yang, Zheng Zhu, Hexiang Wei, Yuntian Zhao, Jing Wu, Zhenzhong Jia
Terrain-aware perception holds the potential to improve the robustness and accuracy of autonomous robot navigation in the wilds, thereby facilitating effective off-road traversals. However, the lack of multi-modal perception across various motion patterns hinders the solutions of Simultaneous Localization And Mapping (SLAM), especially when confronting non-geometric hazards in demanding landscapes. In this paper, we first propose a Terrain-Aware multI-modaL (TAIL) dataset tailored to deformable and sandy terrains. It incorporates various types of robotic proprioception and distinct ground interactions for the unique challenges and benchmark of multi-sensor fusion SLAM. The versatile sensor suite comprises stereo frame cameras, multiple ground-pointing RGB-D cameras, a rotating 3D LiDAR, an IMU, and an RTK device. This ensemble is hardware-synchronized, well-calibrated, and self-contained. Utilizing both wheeled and quadrupedal locomotion, we efficiently collect comprehensive sequences to capture rich unstructured scenarios. It spans the spectrum of scope, terrain interactions, scene changes, ground-level properties, and dynamic robot characteristics. We benchmark several state-of-the-art SLAM methods against ground truth and provide performance validations. Corresponding challenges and limitations are also reported. All associated resources are accessible upon request at \url{https://tailrobot.github.io/}.
Zirui Wang, Chen Yao, Yangtao Ge, Guowei Shi, Ningbo Yang, Zheng Zhu, Kewei Dong, Hexiang Wei, Zhenzhong Jia, Jing Wu
So far, planetary surface exploration depends on various mobile robot platforms. The autonomous navigation and decision-making of these mobile robots in complex terrains largely rely on their terrain-aware perception, localization and mapping capabilities. In this paper we release the TAIL-Plus dataset, a new challenging dataset in deformable granular environments for planetary exploration robots, which is an extension to our previous work, TAIL (Terrain-Aware multI-modaL) dataset. We conducted field experiments on beaches that are considered as planetary surface analog environments for diverse sandy terrains. In TAIL-Plus dataset, we provide more sequences with multiple loops and expand the scene from day to night. Benefit from our sensor suite with modular design, we use both wheeled and quadruped robots for data collection. The sensors include a 3D LiDAR, three downward RGB-D cameras, a pair of global-shutter color cameras that can be used as a forward-looking stereo camera, an RTK-GPS device and an extra IMU. Our datasets are intended to help researchers developing multi-sensor simultaneous localization and mapping (SLAM) algorithms for robots in unstructured, deformable granular terrains. Our datasets and supplementary materials will be available at \url{https://tailrobot.github.io/}.
Zirui Wang, Tao Zhang
3D understanding is a key capability for real-world AI assistance. High-quality data plays an important role in driving the development of the 3D understanding community. Current 3D scene understanding datasets often provide geometric and instance-level information, yet they lack the rich semantic annotations necessary for nuanced visual-language tasks.In this work, we introduce DenseScan, a novel dataset with detailed multi-level descriptions generated by an automated pipeline leveraging multi-view 2D images and multimodal large language models (MLLMs). Our approach enables dense captioning of scene elements, ensuring comprehensive object-level descriptions that capture context-sensitive details. Furthermore, we extend these annotations through scenario-based question generation, producing high-level queries that integrate object properties, spatial relationships, and scene context. By coupling geometric detail with semantic richness, DenseScan broadens the range of downstream tasks, from detailed visual-language navigation to interactive question answering. Experimental results demonstrate that our method significantly enhances object-level understanding and question-answering performance in 3D environments compared to traditional annotation pipelines. We release both the annotated dataset and our annotation pipeline to facilitate future research and applications in robotics, augmented reality, and beyond. Through DenseScan, we aim to catalyze new avenues in 3D scene understanding, allowing researchers and practitioners to tackle the complexities of real-world environments with richer, more contextually aware annotations.
Zirui Wang, Sanket Vaibhav Mehta, Barnabás Póczos, Jaime Carbonell
Current natural language processing models work well on a single task, yet they often fail to continuously learn new tasks without forgetting previous ones as they are re-trained throughout their lifetime, a challenge known as lifelong learning. State-of-the-art lifelong language learning methods store past examples in episodic memory and replay them at both training and inference time. However, as we show later in our experiments, there are three significant impediments: (1) needing unrealistically large memory module to achieve good performance, (2) suffering from negative transfer, (3) requiring multiple local adaptation steps for each test example that significantly slows down the inference speed. In this paper, we identify three common principles of lifelong learning methods and propose an efficient meta-lifelong framework that combines them in a synergistic fashion. To achieve sample efficiency, our method trains the model in a manner that it learns a better initialization for local adaptation. Extensive experiments on text classification and question answering benchmarks demonstrate the effectiveness of our framework by achieving state-of-the-art performance using merely 1% memory size and narrowing the gap with multi-task learning. We further show that our method alleviates both catastrophic forgetting and negative transfer at the same time.
Zirui Wang, Victor Adrian Prisacariu
We introduce a novel self-attention-based normal estimation network that is able to focus softly on relevant points and adjust the softness by learning a temperature parameter, making it able to work naturally and effectively within a large neighbourhood range. As a result, our model outperforms all existing normal estimation algorithms by a large margin, achieving 94.1% accuracy in comparison with the previous state of the art of 91.2%, with a 25x smaller model and 12x faster inference time. We also use point-to-plane Iterative Closest Point (ICP) as an application case to show that our normal estimations lead to faster convergence than normal estimations from other methods, without manually fine-tuning neighbourhood range parameters. Code available at https://code.active.vision.
Zirui Wang, Shangzhe Wu, Weidi Xie, Min Chen, Victor Adrian Prisacariu
Considering the problem of novel view synthesis (NVS) from only a set of 2D images, we simplify the training process of Neural Radiance Field (NeRF) on forward-facing scenes by removing the requirement of known or pre-computed camera parameters, including both intrinsics and 6DoF poses. To this end, we propose NeRF$--$, with three contributions: First, we show that the camera parameters can be jointly optimised as learnable parameters with NeRF training, through a photometric reconstruction; Second, to benchmark the camera parameter estimation and the quality of novel view renderings, we introduce a new dataset of path-traced synthetic scenes, termed as Blender Forward-Facing Dataset (BLEFF); Third, we conduct extensive analyses to understand the training behaviours under various camera motions, and show that in most scenarios, the joint optimisation pipeline can recover accurate camera parameters and achieve comparable novel view synthesis quality as those trained with COLMAP pre-computed camera parameters. Our code and data are available at https://nerfmm.active.vision.
Zirui Wang, Wenjing Bian, Victor Adrian Prisacariu
We introduce a novel cross-reference image quality assessment method that effectively fills the gap in the image assessment landscape, complementing the array of established evaluation schemes -- ranging from full-reference metrics like SSIM, no-reference metrics such as NIQE, to general-reference metrics including FID, and Multi-modal-reference metrics, e.g., CLIPScore. Utilising a neural network with the cross-attention mechanism and a unique data collection pipeline from NVS optimisation, our method enables accurate image quality assessment without requiring ground truth references. By comparing a query image against multiple views of the same scene, our method addresses the limitations of existing metrics in novel view synthesis (NVS) and similar tasks where direct reference images are unavailable. Experimental results show that our method is closely correlated to the full-reference metric SSIM, while not requiring ground truth references.