Yuxin Guo

Maurus Item, Juan Gómez-Luna, Yuxin Guo et al.

Processing-in-memory (PIM) promises to alleviate the data movement bottleneck in modern computing systems. However, current real-world PIM systems have the inherent disadvantage that their hardware is more constrained than in conventional processors (CPU, GPU), due to the difficulty and cost of building processing elements near or inside the memory. As a result, general-purpose PIM architectures support fairly limited instruction sets and struggle to execute complex operations such as transcendental functions and other hard-to-calculate operations (e.g., square root). These operations are particularly important for some modern workloads, e.g., activation functions in machine learning applications. In order to provide support for transcendental (and other hard-to-calculate) functions in general-purpose PIM systems, we present \emph{TransPimLib}, a library that provides CORDIC-based and LUT-based methods for trigonometric functions, hyperbolic functions, exponentiation, logarithm, square root, etc. We develop an implementation of TransPimLib for the UPMEM PIM architecture and perform a thorough evaluation of TransPimLib's methods in terms of performance and accuracy, using microbenchmarks and three full workloads (Blackscholes, Sigmoid, Softmax). We open-source all our code and datasets at~\url{https://github.com/CMU-SAFARI/transpimlib}.

Zi Lin, Zihan Wang, Yongqi Tong et al.

Despite remarkable advances that large language models have achieved in chatbots, maintaining a non-toxic user-AI interactive environment has become increasingly critical nowadays. However, previous efforts in toxicity detection have been mostly based on benchmarks derived from social media content, leaving the unique challenges inherent to real-world user-AI interactions insufficiently explored. In this work, we introduce ToxicChat, a novel benchmark based on real user queries from an open-source chatbot. This benchmark contains the rich, nuanced phenomena that can be tricky for current toxicity detection models to identify, revealing a significant domain difference compared to social media content. Our systematic evaluation of models trained on existing toxicity datasets has shown their shortcomings when applied to this unique domain of ToxicChat. Our work illuminates the potentially overlooked challenges of toxicity detection in real-world user-AI conversations. In the future, ToxicChat can be a valuable resource to drive further advancements toward building a safe and healthy environment for user-AI interactions.

Juan Gómez-Luna, Yuxin Guo, Sylvan Brocard et al.

Training machine learning (ML) algorithms is a computationally intensive process, which is frequently memory-bound due to repeatedly accessing large training datasets. As a result, processor-centric systems (e.g., CPU, GPU) suffer from costly data movement between memory units and processing units, which consumes large amounts of energy and execution cycles. Memory-centric computing systems, i.e., with processing-in-memory (PIM) capabilities, can alleviate this data movement bottleneck. Our goal is to understand the potential of modern general-purpose PIM architectures to accelerate ML training. To do so, we (1) implement several representative classic ML algorithms (namely, linear regression, logistic regression, decision tree, K-Means clustering) on a real-world general-purpose PIM architecture, (2) rigorously evaluate and characterize them in terms of accuracy, performance and scaling, and (3) compare to their counterpart implementations on CPU and GPU. Our evaluation on a real memory-centric computing system with more than 2500 PIM cores shows that general-purpose PIM architectures can greatly accelerate memory-bound ML workloads, when the necessary operations and datatypes are natively supported by PIM hardware. For example, our PIM implementation of decision tree is $27\times$ faster than a state-of-the-art CPU version on an 8-core Intel Xeon, and $1.34\times$ faster than a state-of-the-art GPU version on an NVIDIA A100. Our K-Means clustering on PIM is $2.8\times$ and $3.2\times$ than state-of-the-art CPU and GPU versions, respectively. To our knowledge, our work is the first one to evaluate ML training on a real-world PIM architecture. We conclude with key observations, takeaways, and recommendations that can inspire users of ML workloads, programmers of PIM architectures, and hardware designers & architects of future memory-centric computing systems.

Hengyu Mu, Jianshi Wu, Yuxin Guo et al.

Regression-based LiDAR relocalization has recently emerged as a promising solution for high-precision positioning in GNSS-denied environments. However, these methods are primarily tailored to autonomous driving, exhibiting significantly degraded accuracy in unmanned aerial vehicle (UAV) scenarios due to arbitrary pose variations and irregular flight paths. In this paper, we propose SOAR, a regression-based LiDAR relocalization framework for UAVs. Specifically, we introduce a locality-preserving sliding window attention module with locally invariant positional encoding to capture discriminative geometric structures robust to viewpoint changes. A coordinate-independent feature initialization module is further designed to eliminate sensitivity to global transformations. Furthermore, most existing UAV datasets are limited to evaluate LiDAR relocalization in real-world, due to the lack of synchronized LiDAR scans, accurate 6-DoF poses, or multiple traversals. Thus, we construct a large-scale UAV LiDAR localization dataset with 4 scenes and 13 irregular paths exhibiting rotation and altitude variations, providing a more realistic benchmark for UAVs. Extensive experiments demonstrate that our method achieves state-of-the-art performance, improving the localization success rate by 40% and reducing mean error over 10m on UAVLoc. Our code and dataset will be released soon.

Juan Gómez-Luna, Yuxin Guo, Sylvan Brocard et al.

Training machine learning algorithms is a computationally intensive process, which is frequently memory-bound due to repeatedly accessing large training datasets. As a result, processor-centric systems (e.g., CPU, GPU) suffer from costly data movement between memory units and processing units, which consumes large amounts of energy and execution cycles. Memory-centric computing systems, i.e., computing systems with processing-in-memory (PIM) capabilities, can alleviate this data movement bottleneck. Our goal is to understand the potential of modern general-purpose PIM architectures to accelerate machine learning training. To do so, we (1) implement several representative classic machine learning algorithms (namely, linear regression, logistic regression, decision tree, K-means clustering) on a real-world general-purpose PIM architecture, (2) characterize them in terms of accuracy, performance and scaling, and (3) compare to their counterpart implementations on CPU and GPU. Our experimental evaluation on a memory-centric computing system with more than 2500 PIM cores shows that general-purpose PIM architectures can greatly accelerate memory-bound machine learning workloads, when the necessary operations and datatypes are natively supported by PIM hardware. To our knowledge, our work is the first one to evaluate training of machine learning algorithms on a real-world general-purpose PIM architecture.

Yuxin Guo, Deyu Bo, Cheng Yang et al.

The history of artificial intelligence (AI) has witnessed the significant impact of high-quality data on various deep learning models, such as ImageNet for AlexNet and ResNet. Recently, instead of designing more complex neural architectures as model-centric approaches, the attention of AI community has shifted to data-centric ones, which focuses on better processing data to strengthen the ability of neural models. Graph learning, which operates on ubiquitous topological data, also plays an important role in the era of deep learning. In this survey, we comprehensively review graph learning approaches from the data-centric perspective, and aim to answer three crucial questions: (1) when to modify graph data, (2) what part of the graph data needs modification to unlock the potential of various graph models, and (3) how to safeguard graph models from problematic data influence. Accordingly, we propose a novel taxonomy based on the stages in the graph learning pipeline, and highlight the processing methods for different data structures in the graph data, i.e., topology, feature and label. Furthermore, we analyze some potential problems embedded in graph data and discuss how to solve them in a data-centric manner. Finally, we provide some promising future directions for data-centric graph learning.

Yue Fang, Weibin Liao, Yuxin Guo et al.

Clinical Reasoning on Electronic Health Records (EHRs) is a fundamental yet challenging task in modern healthcare. While in-context learning (ICL) offers a promising inference-time adaptation paradigm for large language models (LLMs) in EHR reasoning, existing methods face three fundamental challenges: (1) Perspective Limitation, where data-driven similarity fails to align with LLM reasoning needs and model-driven signals are constrained by limited clinical competence; (2) Cohort Awareness, as demonstrations are selected independently without modeling population-level structure; and (3) Information Aggregation, where redundancy and interaction effects among demonstrations are ignored, leading to diminishing marginal gains. To address these challenges, we propose GraphWalker, a principled demonstration selection framework for EHR-oriented ICL. GraphWalker (i) jointly models patient clinical information and LLM-estimated information gain by integrating data-driven and model-driven perspectives, (ii) incorporates Cohort Discovery to avoid noisy local optima, and (iii) employs a Lazy Greedy Search with Frontier Expansion algorithm to mitigate diminishing marginal returns in information aggregation. Extensive experiments on multiple real-world EHR benchmarks demonstrate that GraphWalker consistently outperforms state-of-the-art ICL baselines, yielding substantial improvements in clinical reasoning performance. Our code is open-sourced at https://github.com/PuppyKnightUniversity/GraphWalker

Chunyu Qiang, Xiaopeng Wang, Kang Yin et al.

Generative audio modeling has largely been fragmented into specialized tasks, text-to-speech (TTS), text-to-music (TTM), and text-to-audio (TTA), each operating under heterogeneous control paradigms. Unifying these modalities remains a fundamental challenge due to the intrinsic dissonance between structured semantic representations (speech/music) and unstructured acoustic textures (sound effects). In this paper, we introduce UniSonate, a unified flow-matching framework capable of synthesizing speech, music, and sound effects through a standardized, reference-free natural language instruction interface. To reconcile structural disparities, we propose a novel dynamic token injection mechanism that projects unstructured environmental sounds into a structured temporal latent space, enabling precise duration control within a phoneme-driven Multimodal Diffusion Transformer (MM-DiT). Coupled with a multi-stage curriculum learning strategy, this approach effectively mitigates cross-modal optimization conflicts. Extensive experiments demonstrate that UniSonate achieves state-of-the-art performance in instruction-based TTS (WER 1.47%) and TTM (SongEval Coherence 3.18), while maintaining competitive fidelity in TTA. Crucially, we observe positive transfer, where joint training on diverse audio data significantly enhances structural coherence and prosodic expressiveness compared to single-task baselines. Audio samples are available at https://qiangchunyu.github.io/UniSonate/.

Yuxin Guo, Yihao Yue, Yunhao Ni et al.

Layer normalization (LN) is a fundamental component in modern deep learning, but its per-sample centering and scaling introduce non-negligible inference overhead. RMSNorm improves efficiency by removing the centering operation, yet this may discard benefits associated with centering. This paper propose a framework to determine whether an LN in an arbitrary DNN can be replaced by RMSNorm without changing the model function. The key idea is to fold LN's centering operation into upstream general linear layers by enforcing zero-mean outputs through the column-centered constraint (CCC) and column-based weight centering (CBWC). We extend the analysis to arbitrary DNNs, define such LNs as foldable LNs, and develop a graph-based detection algorithm. Our analysis shows that many LNs in widely used architectures are foldable, enabling exact inference-time conversion and end-to-end acceleration of 2% to 12% without changing model predictions. Experiments across multiple task families further show that, when exact equivalence is partially broken in practical training settings, our method remains competitive with vanilla LN while improving efficiency.

Yuxin Guo, Dongrui Deng, Pulkit Grover

Recently, deep generative models have been used for posterior inference in inverse problems, including high-stakes applications in medical imaging and scientific discovery, where the uncertainty of a prediction can matter as much as the prediction itself. However, posterior uncertainty is difficult to interpret because it can mix ambiguity inherent to the forward operator with uncertainty propagated through inference. We introduce a structural decomposition of posterior uncertainty that isolates intrinsic ambiguity. A cascade formulation makes this ambiguity accessible for calibration analysis, enabling qualitative diagnostics and simulation-based calibration tests that reveal failure modes that remain hidden when models are selected by reconstruction quality alone. We first validate the approach on a Gaussian example with analytical posterior structure, then illustrate the decomposition on accelerated magnetic resonance imaging (MRI), and finally apply the calibration diagnostics to electroencephalography (EEG) source imaging.

Yue Fang, Yuxin Guo, Jiaran Gao et al.

Improving large language models (LLMs) for electronic health record (EHR) reasoning is essential for enabling accurate and generalizable clinical predictions. While LLMs excel at medical text understanding, they underperform on EHR-based prediction tasks due to challenges in modeling temporally structured, high-dimensional data. Existing approaches often rely on hybrid paradigms, where LLMs serve merely as frozen prior retrievers while downstream deep learning (DL) models handle prediction, failing to improve the LLM's intrinsic reasoning capacity and inheriting the generalization limitations of DL models. To this end, we propose EAG-RL, a novel two-stage training framework designed to intrinsically enhance LLMs' EHR reasoning ability through expert attention guidance, where expert EHR models refer to task-specific DL models trained on EHR data. Concretely, EAG-RL first constructs high-quality, stepwise reasoning trajectories using expert-guided Monte Carlo Tree Search to effectively initialize the LLM's policy. Then, EAG-RL further optimizes the policy via reinforcement learning by aligning the LLM's attention with clinically salient features identified by expert EHR models. Extensive experiments on two real-world EHR datasets show that EAG-RL improves the intrinsic EHR reasoning ability of LLMs by an average of 14.62%, while also enhancing robustness to feature perturbations and generalization to unseen clinical domains. These results demonstrate the practical potential of EAG-RL for real-world deployment in clinical prediction tasks. Our code have been available at https://github.com/devilran6/EAG-RL.

Yuxin Guo, Teng Wang, Yuying Ge et al.

Recent advances in text-to-audio (TTA) generation excel at synthesizing short audio clips but struggle with long-form narrative audio, which requires temporal coherence and compositional reasoning. To address this gap, we propose AudioStory, a unified framework that integrates large language models (LLMs) with TTA systems to generate structured, long-form audio narratives. AudioStory possesses strong instruction-following reasoning generation capabilities. It employs LLMs to decompose complex narrative queries into temporally ordered sub-tasks with contextual cues, enabling coherent scene transitions and emotional tone consistency. AudioStory has two appealing features: (1) Decoupled bridging mechanism: AudioStory disentangles LLM-diffuser collaboration into two specialized components, i.e., a bridging query for intra-event semantic alignment and a residual query for cross-event coherence preservation. (2) End-to-end training: By unifying instruction comprehension and audio generation within a single end-to-end framework, AudioStory eliminates the need for modular training pipelines while enhancing synergy between components. Furthermore, we establish a benchmark AudioStory-10K, encompassing diverse domains such as animated soundscapes and natural sound narratives. Extensive experiments show the superiority of AudioStory on both single-audio generation and narrative audio generation, surpassing prior TTA baselines in both instruction-following ability and audio fidelity. Our code is available at https://github.com/TencentARC/AudioStory

Jiangyuan Liu, Hongxuan Ma, Yuxin Guo et al.

Transparent object perception is indispensable for numerous robotic tasks. However, accurately segmenting and estimating the depth of transparent objects remain challenging due to complex optical properties. Existing methods primarily delve into only one task using extra inputs or specialized sensors, neglecting the valuable interactions among tasks and the subsequent refinement process, leading to suboptimal and blurry predictions. To address these issues, we propose a monocular framework, which is the first to excel in both segmentation and depth estimation of transparent objects, with only a single-image input. Specifically, we devise a novel semantic and geometric fusion module, effectively integrating the multi-scale information between tasks. In addition, drawing inspiration from human perception of objects, we further incorporate an iterative strategy, which progressively refines initial features for clearer results. Experiments on two challenging synthetic and real-world datasets demonstrate that our model surpasses state-of-the-art monocular, stereo, and multi-view methods by a large margin of about 38.8%-46.2% with only a single RGB input. Codes and models are publicly available at https://github.com/L-J-Yuan/MODEST.

Jun Wang, Chunyu Qiang, Yuxin Guo et al.

Audio-video joint generation has progressed rapidly, yet substantial challenges still remain. Non-commercial approaches still suffer audio-visual asynchrony, poor lip-speech alignment, and unimodal degradation, which can be stemmed from weak audio-visual correspondence modeling, limited generalization, and scarce high-quality dense-caption data. To address these issues, we introduce Klear and delve into three axes--model architecture, training strategy, and data curation. Architecturally, we adopt a single-tower design with unified DiT blocks and an Omni-Full Attention mechanism, achieving tight audio-visual alignment and strong scalability. Training-wise, we adopt a progressive multitask regime--random modality masking to joint optimization across tasks, and a multistage curriculum, yielding robust representations, strengthening A-V aligned world knowledge, and preventing unimodal collapse. For datasets, we present the first large-scale audio-video dataset with dense captions, and introduce a novel automated data-construction pipeline which annotates and filters millions of diverse, high-quality, strictly aligned audio-video-caption triplets. Building on this, Klear scales to large datasets, delivering high-fidelity, semantically and temporally aligned, instruction-following generation in both joint and unimodal settings while generalizing robustly to out-of-distribution scenarios. Across tasks, it substantially outperforms prior methods by a large margin and achieves performance comparable to Veo 3, offering a unified, scalable path toward next-generation audio-video synthesis.

Ziqian Wang, Yonghao He, Licheng Yang et al.

Generating high-fidelity, physically interactive 3D simulated tabletop scenes is essential for embodied AI--especially for robotic manipulation policy learning and data synthesis. However, current text- or image-driven 3D scene generation methods mainly focus on large-scale scenes, struggling to capture the high-density layouts and complex spatial relations that characterize tabletop scenes. To address these challenges, we propose TabletopGen, a training-free, fully automatic framework that generates diverse, instance-level interactive 3D tabletop scenes. TabletopGen accepts a reference image as input, which can be synthesized by a text-to-image model to enhance scene diversity. We then perform instance segmentation and completion on the reference to obtain per-instance images. Each instance is reconstructed into a 3D model followed by canonical coordinate alignment. The aligned 3D models then undergo pose and scale estimation before being assembled into a collision-free, simulation-ready tabletop scene. A key component of our framework is a novel pose and scale alignment approach that decouples the complex spatial reasoning into two stages: a Differentiable Rotation Optimizer for precise rotation recovery and a Top-view Spatial Alignment mechanism for robust translation and scale estimation, enabling accurate 3D reconstruction from 2D reference. Extensive experiments and user studies show that TabletopGen achieves state-of-the-art performance, markedly surpassing existing methods in visual fidelity, layout accuracy, and physical plausibility, capable of generating realistic tabletop scenes with rich stylistic and spatial diversity. Our code will be publicly available.

Zehuan Wu, Jingcheng Ni, Xiaodong Wang et al.

Generative models have significantly improved the generation and prediction quality on either camera images or LiDAR point clouds for autonomous driving. However, a real-world autonomous driving system uses multiple kinds of input modality, usually cameras and LiDARs, where they contain complementary information for generation, while existing generation methods ignore this crucial feature, resulting in the generated results only covering separate 2D or 3D information. In order to fill the gap in 2D-3D multi-modal joint generation for autonomous driving, in this paper, we propose our framework, \emph{HoloDrive}, to jointly generate the camera images and LiDAR point clouds. We employ BEV-to-Camera and Camera-to-BEV transform modules between heterogeneous generative models, and introduce a depth prediction branch in the 2D generative model to disambiguate the un-projecting from image space to BEV space, then extend the method to predict the future by adding temporal structure and carefully designed progressive training. Further, we conduct experiments on single frame generation and world model benchmarks, and demonstrate our method leads to significant performance gains over SOTA methods in terms of generation metrics.

Yuxin Guo, Shijie Ma, Hu Su et al.

Audio-Visual Source Localization (AVSL) aims to locate sounding objects within video frames given the paired audio clips. Existing methods predominantly rely on self-supervised contrastive learning of audio-visual correspondence. Without any bounding-box annotations, they struggle to achieve precise localization, especially for small objects, and suffer from blurry boundaries and false positives. Moreover, the naive semi-supervised method is poor in fully leveraging the information of abundant unlabeled data. In this paper, we propose a novel semi-supervised learning framework for AVSL, namely Dual Mean-Teacher (DMT), comprising two teacher-student structures to circumvent the confirmation bias issue. Specifically, two teachers, pre-trained on limited labeled data, are employed to filter out noisy samples via the consensus between their predictions, and then generate high-quality pseudo-labels by intersecting their confidence maps. The sufficient utilization of both labeled and unlabeled data and the proposed unbiased framework enable DMT to outperform current state-of-the-art methods by a large margin, with CIoU of 90.4% and 48.8% on Flickr-SoundNet and VGG-Sound Source, obtaining 8.9%, 9.6% and 4.6%, 6.4% improvements over self- and semi-supervised methods respectively, given only 3% positional-annotations. We also extend our framework to some existing AVSL methods and consistently boost their performance.

Xiaoyi Bao, Siyang Sun, Shuailei Ma et al.

The reasoning segmentation task, which demands a nuanced comprehension of intricate queries to accurately pinpoint object regions, is attracting increasing attention. However, Multi-modal Large Language Models (MLLM) often find it difficult to accurately localize the objects described in complex reasoning contexts. We believe that the act of reasoning segmentation should mirror the cognitive stages of human visual search, where each step is a progressive refinement of thought toward the final object. Thus we introduce the Chains of Reasoning and Segmenting (CoReS) and find this top-down visual hierarchy indeed enhances the visual search process. Specifically, we propose a dual-chain structure that generates multi-modal, chain-like outputs to aid the segmentation process. Furthermore, to steer the MLLM's outputs into this intended hierarchy, we incorporate in-context inputs as guidance. Extensive experiments demonstrate the superior performance of our CoReS, which surpasses the state-of-the-art method by 6.5\% on the ReasonSeg dataset. Project: https://chain-of-reasoning-and-segmentation.github.io/.

Jingcheng Ni, Yuxin Guo, Yichen Liu et al.

World models that forecast environmental changes from actions are vital for autonomous driving models with strong generalization. The prevailing driving world model mainly build on video prediction model. Although these models can produce high-fidelity video sequences with advanced diffusion-based generator, they are constrained by their predictive duration and overall generalization capabilities. In this paper, we explore to solve this problem by combining generation loss with MAE-style feature-level context learning. In particular, we instantiate this target with three key design: (1) A more scalable Diffusion Transformer (DiT) structure trained with extra mask construction task. (2) we devise diffusion-related mask tokens to deal with the fuzzy relations between mask reconstruction and generative diffusion process. (3) we extend mask construction task to spatial-temporal domain by utilizing row-wise mask for shifted self-attention rather than masked self-attention in MAE. Then, we adopt a row-wise cross-view module to align with this mask design. Based on above improvement, we propose MaskGWM: a Generalizable driving World Model embodied with Video Mask reconstruction. Our model contains two variants: MaskGWM-long, focusing on long-horizon prediction, and MaskGWM-mview, dedicated to multi-view generation. Comprehensive experiments on standard benchmarks validate the effectiveness of the proposed method, which contain normal validation of Nuscene dataset, long-horizon rollout of OpenDV-2K dataset and zero-shot validation of Waymo dataset. Quantitative metrics on these datasets show our method notably improving state-of-the-art driving world model.

Rui Chen, Zehuan Wu, Yichen Liu et al.

The creation of diverse and realistic driving scenarios has become essential to enhance perception and planning capabilities of the autonomous driving system. However, generating long-duration, surround-view consistent driving videos remains a significant challenge. To address this, we present UniMLVG, a unified framework designed to generate extended street multi-perspective videos under precise control. By integrating single- and multi-view driving videos into the training data, our approach updates a DiT-based diffusion model equipped with cross-frame and cross-view modules across three stages with multi training objectives, substantially boosting the diversity and quality of generated visual content. Importantly, we propose an innovative explicit viewpoint modeling approach for multi-view video generation to effectively improve motion transition consistency. Capable of handling various input reference formats (e.g., text, images, or video), our UniMLVG generates high-quality multi-view videos according to the corresponding condition constraints such as 3D bounding boxes or frame-level text descriptions. Compared to the best models with similar capabilities, our framework achieves improvements of 48.2% in FID and 35.2% in FVD.

Shijie Ma, Yuying Ge, Teng Wang et al.

The synergy between generative and discriminative models receives growing attention. While discriminative Contrastive Language-Image Pre-Training (CLIP) excels in high-level semantics, it struggles with perceiving fine-grained visual details. Generally, to enhance representations, generative models take CLIP's visual features as conditions for reconstruction. However, the underlying principle remains underexplored. In this work, we empirically found that visually perfect generations are not always optimal for representation enhancement. The essence lies in effectively extracting fine-grained knowledge from generative models while mitigating irrelevant information. To explore critical factors, we delve into three aspects: (1) Conditioning mechanisms: We found that even a small number of local tokens can drastically reduce the difficulty of reconstruction, leading to collapsed training. We thus conclude that utilizing only global visual tokens as conditions is the most effective strategy. (2) Denoising configurations: We observed that end-to-end training introduces extraneous information. To address this, we propose a two-stage training strategy to prioritize learning useful visual knowledge. Additionally, we demonstrate that lightweight denoisers can yield remarkable improvements. (3) Generation paradigms: We explore both continuous and discrete denoisers with desirable outcomes, validating the versatility of our method. Through our in-depth explorations, we have finally arrived at an effective method, namely GenHancer, which consistently outperforms prior arts on the MMVP-VLM benchmark, e.g., 6.0% on OpenAICLIP. The enhanced CLIP can be further plugged into multimodal large language models for better vision-centric performance. All the models and codes are made publicly available.

Minghang Zhu, Zhijing Wang, Yuxin Guo et al.

LiDAR relocalization aims to estimate the global 6-DoF pose of a sensor in the environment. However, existing regression-based approaches are prone to dynamic or ambiguous scenarios, as they either solely rely on single-frame inference or neglect the spatio-temporal consistency across scans. In this paper, we propose TempLoc, a new LiDAR relocalization framework that enhances the robustness of localization by effectively modeling sequential consistency. Specifically, a Global Coordinate Estimation module is first introduced to predict point-wise global coordinates and associated uncertainties for each LiDAR scan. A Prior Coordinate Generation module is then presented to estimate inter-frame point correspondences by the attention mechanism. Lastly, an Uncertainty-Guided Coordinate Fusion module is deployed to integrate both predictions of point correspondence in an end-to-end fashion, yielding a more temporally consistent and accurate global 6-DoF pose. Experimental results on the NCLT and Oxford Robot-Car benchmarks show that our TempLoc outperforms stateof-the-art methods by a large margin, demonstrating the effectiveness of temporal-aware correspondence modeling in LiDAR relocalization. Our code will be released soon.

Yuxin Guo, Shuailei Ma, Shijie Ma et al.

Audio is essential for multimodal video understanding. On the one hand, video inherently contains audio, which supplies complementary information to vision. Besides, video large language models (Video-LLMs) can encounter many audio-centric settings. However, existing Video-LLMs and Audio-Visual Large Language Models (AV-LLMs) exhibit deficiencies in exploiting audio information, leading to weak understanding and hallucinations. To solve the issues, we delve into the model architecture and dataset. (1) From the architectural perspective, we propose a fine-grained AV-LLM, namely Dolphin. The concurrent alignment of audio and visual modalities in both temporal and spatial dimensions ensures a comprehensive and accurate understanding of videos. Specifically, we devise an audio-visual multi-scale adapter for multi-scale information aggregation, which achieves spatial alignment. For temporal alignment, we propose audio-visual interleaved merging. (2) From the dataset perspective, we curate an audio-visual caption and instruction-tuning dataset, called AVU. It comprises 5.2 million diverse, open-ended data tuples (video, audio, question, answer) and introduces a novel data partitioning strategy. Extensive experiments show our model not only achieves remarkable performance in audio-visual understanding, but also mitigates potential hallucinations.

Yuxin Guo, Shijie Ma, Yuhao Zhao et al.

Audio-Visual Source Localization (AVSL) is the task of identifying specific sounding objects in the scene given audio cues. In our work, we focus on semi-supervised AVSL with pseudo-labeling. To address the issues with vanilla hard pseudo-labels including bias accumulation, noise sensitivity, and instability, we propose a novel method named Cross Pseudo-Labeling (XPL), wherein two models learn from each other with the cross-refine mechanism to avoid bias accumulation. We equip XPL with two effective components. Firstly, the soft pseudo-labels with sharpening and pseudo-label exponential moving average mechanisms enable models to achieve gradual self-improvement and ensure stable training. Secondly, the curriculum data selection module adaptively selects pseudo-labels with high quality during training to mitigate potential bias. Experimental results demonstrate that XPL significantly outperforms existing methods, achieving state-of-the-art performance while effectively mitigating confirmation bias and ensuring training stability.

Shuailei Ma, Chen-Wei Xie, Ying Wei et al.

Prompt learning has emerged as an efficient alternative for fine-tuning foundational models, such as CLIP, for various downstream tasks. However, there is no work that provides a comprehensive explanation for the working mechanism of the multi-modal prompts. In this paper, we conduct a direct analysis of the multi-modal prompts by asking the following questions: $(i)$ How do the learned multi-modal prompts improve the recognition performance? $(ii)$ What do the multi-modal prompts learn? To answer these questions, we begin by isolating the component of the formula where the prompt influences the calculation of self-attention at each layer in two distinct ways, \ie, $(1)$ introducing prompt embeddings makes the $[cls]$ token focus on foreground objects. $(2)$ the prompts learn a bias term during the update of token embeddings, allowing the model to adapt to the target domain. Subsequently, we conduct extensive visualization and statistical experiments on the eleven diverse downstream recognition datasets. From the experiments, we reveal that the learned prompts improve the performance mainly through the second way, which acts as the dataset bias to improve the recognition performance of the pre-trained model on the corresponding dataset. Meanwhile, we propose the bias tuning way to validate our finding. With a deeper understanding of the multi-modal prompt, we hope our work can inspire new and solid research in this direction.

Chunyu Qiang, Jun Wang, Xiaopeng Wang et al.

Joint audio-video generation aims to synthesize synchronized multisensory content, yet current unified models struggle with fine-grained acoustic control, particularly for identity-preserving speech. Existing approaches either suffer from temporal misalignment due to cascaded generation or lack the capability to perform zero-shot voice cloning within a joint synthesis framework. In this work, we present MM-Sonate, a multimodal flow-matching framework that unifies controllable audio-video joint generation with zero-shot voice cloning capabilities. Unlike prior works that rely on coarse semantic descriptions, MM-Sonate utilizes a unified instruction-phoneme input to enforce strict linguistic and temporal alignment. To enable zero-shot voice cloning, we introduce a timbre injection mechanism that effectively decouples speaker identity from linguistic content. Furthermore, addressing the limitations of standard classifier-free guidance in multimodal settings, we propose a noise-based negative conditioning strategy that utilizes natural noise priors to significantly enhance acoustic fidelity. Empirical evaluations demonstrate that MM-Sonate establishes new state-of-the-art performance in joint generation benchmarks, significantly outperforming baselines in lip synchronization and speech intelligibility, while achieving voice cloning fidelity comparable to specialized Text-to-Speech systems.

Tianrui Zhang, Yichen Liu, Zilin Guo et al.

Generative models have been widely applied to world modeling for environment simulation and future state prediction. With advancements in autonomous driving, there is a growing demand not only for high-fidelity video generation under various controls, but also for producing diverse and meaningful information such as depth estimation. To address this, we propose CVD-STORM, a cross-view video diffusion model utilizing a spatial-temporal reconstruction Variational Autoencoder (VAE) that generates long-term, multi-view videos with 4D reconstruction capabilities under various control inputs. Our approach first fine-tunes the VAE with an auxiliary 4D reconstruction task, enhancing its ability to encode 3D structures and temporal dynamics. Subsequently, we integrate this VAE into the video diffusion process to significantly improve generation quality. Experimental results demonstrate that our model achieves substantial improvements in both FID and FVD metrics. Additionally, the jointly-trained Gaussian Splatting Decoder effectively reconstructs dynamic scenes, providing valuable geometric information for comprehensive scene understanding. Our project page is https://sensetime-fvg.github.io/CVD-STORM.

Yunhao Ni, Yuhe Liu, Wenxin Sun et al.

Universal approximation theorem (UAT) is a fundamental theory for deep neural networks (DNNs), demonstrating their powerful representation capacity to represent and approximate any function. The analyses and proofs of UAT are based on traditional network with only linear and nonlinear activation functions, but omitting normalization layers, which are commonly employed to enhance the training of modern networks. This paper conducts research on UAT of DNNs with normalization layers for the first time. We theoretically prove that an infinitely wide network -- composed solely of parallel layer normalization (PLN) and linear layers -- has universal approximation capacity. Additionally, we investigate the minimum number of neurons required to approximate $L$-Lipchitz continuous functions, with a single hidden-layer network. We compare the approximation capacity of PLN with traditional activation functions in theory. Different from the traditional activation functions, we identify that PLN can act as both activation function and normalization in deep neural networks at the same time. We also find that PLN can improve the performance when replacing LN in transformer architectures, which reveals the potential of PLN used in neural architectures.

Yunhao Ni, Yuxin Guo, Junlong Jia et al.

Layer normalization (LN) is a ubiquitous technique in deep learning but our theoretical understanding to it remains elusive. This paper investigates a new theoretical direction for LN, regarding to its nonlinearity and representation capacity. We investigate the representation capacity of a network with layerwise composition of linear and LN transformations, referred to as LN-Net. We theoretically show that, given $m$ samples with any label assignment, an LN-Net with only 3 neurons in each layer and $O(m)$ LN layers can correctly classify them. We further show the lower bound of the VC dimension of an LN-Net. The nonlinearity of LN can be amplified by group partition, which is also theoretically demonstrated with mild assumption and empirically supported by our experiments. Based on our analyses, we consider to design neural architecture by exploiting and amplifying the nonlinearity of LN, and the effectiveness is supported by our experiments.