Mingyuan Zhang

Haotong Qin, Mingyuan Zhang, Yifu Ding et al.

Network binarization emerges as one of the most promising compression approaches offering extraordinary computation and memory savings by minimizing the bit-width. However, recent research has shown that applying existing binarization algorithms to diverse tasks, architectures, and hardware in realistic scenarios is still not straightforward. Common challenges of binarization, such as accuracy degradation and efficiency limitation, suggest that its attributes are not fully understood. To close this gap, we present BiBench, a rigorously designed benchmark with in-depth analysis for network binarization. We first carefully scrutinize the requirements of binarization in the actual production and define evaluation tracks and metrics for a comprehensive and fair investigation. Then, we evaluate and analyze a series of milestone binarization algorithms that function at the operator level and with extensive influence. Our benchmark reveals that 1) the binarized operator has a crucial impact on the performance and deployability of binarized networks; 2) the accuracy of binarization varies significantly across different learning tasks and neural architectures; 3) binarization has demonstrated promising efficiency potential on edge devices despite the limited hardware support. The results and analysis also lead to a promising paradigm for accurate and efficient binarization. We believe that BiBench will contribute to the broader adoption of binarization and serve as a foundation for future research. The code for our BiBench is released https://github.com/htqin/BiBench .

Zhongang Cai, Wanqi Yin, Ailing Zeng et al.

Expressive human pose and shape estimation (EHPS) unifies body, hands, and face motion capture with numerous applications. Despite encouraging progress, current state-of-the-art methods still depend largely on a confined set of training datasets. In this work, we investigate scaling up EHPS towards the first generalist foundation model (dubbed SMPLer-X), with up to ViT-Huge as the backbone and training with up to 4.5M instances from diverse data sources. With big data and the large model, SMPLer-X exhibits strong performance across diverse test benchmarks and excellent transferability to even unseen environments. 1) For the data scaling, we perform a systematic investigation on 32 EHPS datasets, including a wide range of scenarios that a model trained on any single dataset cannot handle. More importantly, capitalizing on insights obtained from the extensive benchmarking process, we optimize our training scheme and select datasets that lead to a significant leap in EHPS capabilities. 2) For the model scaling, we take advantage of vision transformers to study the scaling law of model sizes in EHPS. Moreover, our finetuning strategy turn SMPLer-X into specialist models, allowing them to achieve further performance boosts. Notably, our foundation model SMPLer-X consistently delivers state-of-the-art results on seven benchmarks such as AGORA (107.2 mm NMVE), UBody (57.4 mm PVE), EgoBody (63.6 mm PVE), and EHF (62.3 mm PVE without finetuning). Homepage: https://caizhongang.github.io/projects/SMPLer-X/

Zhongang Cai, Daxuan Ren, Ailing Zeng et al.

4D human sensing and modeling are fundamental tasks in vision and graphics with numerous applications. With the advances of new sensors and algorithms, there is an increasing demand for more versatile datasets. In this work, we contribute HuMMan, a large-scale multi-modal 4D human dataset with 1000 human subjects, 400k sequences and 60M frames. HuMMan has several appealing properties: 1) multi-modal data and annotations including color images, point clouds, keypoints, SMPL parameters, and textured meshes; 2) popular mobile device is included in the sensor suite; 3) a set of 500 actions, designed to cover fundamental movements; 4) multiple tasks such as action recognition, pose estimation, parametric human recovery, and textured mesh reconstruction are supported and evaluated. Extensive experiments on HuMMan voice the need for further study on challenges such as fine-grained action recognition, dynamic human mesh reconstruction, point cloud-based parametric human recovery, and cross-device domain gaps.

Mingyuan Zhang, Zhongang Cai, Liang Pan et al.

Human motion modeling is important for many modern graphics applications, which typically require professional skills. In order to remove the skill barriers for laymen, recent motion generation methods can directly generate human motions conditioned on natural languages. However, it remains challenging to achieve diverse and fine-grained motion generation with various text inputs. To address this problem, we propose MotionDiffuse, the first diffusion model-based text-driven motion generation framework, which demonstrates several desired properties over existing methods. 1) Probabilistic Mapping. Instead of a deterministic language-motion mapping, MotionDiffuse generates motions through a series of denoising steps in which variations are injected. 2) Realistic Synthesis. MotionDiffuse excels at modeling complicated data distribution and generating vivid motion sequences. 3) Multi-Level Manipulation. MotionDiffuse responds to fine-grained instructions on body parts, and arbitrary-length motion synthesis with time-varied text prompts. Our experiments show MotionDiffuse outperforms existing SoTA methods by convincing margins on text-driven motion generation and action-conditioned motion generation. A qualitative analysis further demonstrates MotionDiffuse's controllability for comprehensive motion generation. Homepage: https://mingyuan-zhang.github.io/projects/MotionDiffuse.html

Haotong Qin, Yifu Ding, Mingyuan Zhang et al.

The large pre-trained BERT has achieved remarkable performance on Natural Language Processing (NLP) tasks but is also computation and memory expensive. As one of the powerful compression approaches, binarization extremely reduces the computation and memory consumption by utilizing 1-bit parameters and bitwise operations. Unfortunately, the full binarization of BERT (i.e., 1-bit weight, embedding, and activation) usually suffer a significant performance drop, and there is rare study addressing this problem. In this paper, with the theoretical justification and empirical analysis, we identify that the severe performance drop can be mainly attributed to the information degradation and optimization direction mismatch respectively in the forward and backward propagation, and propose BiBERT, an accurate fully binarized BERT, to eliminate the performance bottlenecks. Specifically, BiBERT introduces an efficient Bi-Attention structure for maximizing representation information statistically and a Direction-Matching Distillation (DMD) scheme to optimize the full binarized BERT accurately. Extensive experiments show that BiBERT outperforms both the straightforward baseline and existing state-of-the-art quantized BERTs with ultra-low bit activations by convincing margins on the NLP benchmark. As the first fully binarized BERT, our method yields impressive 56.3 times and 31.2 times saving on FLOPs and model size, demonstrating the vast advantages and potential of the fully binarized BERT model in real-world resource-constrained scenarios.

Mingyuan Zhang, Xinying Guo, Liang Pan et al.

3D human motion generation is crucial for creative industry. Recent advances rely on generative models with domain knowledge for text-driven motion generation, leading to substantial progress in capturing common motions. However, the performance on more diverse motions remains unsatisfactory. In this work, we propose ReMoDiffuse, a diffusion-model-based motion generation framework that integrates a retrieval mechanism to refine the denoising process. ReMoDiffuse enhances the generalizability and diversity of text-driven motion generation with three key designs: 1) Hybrid Retrieval finds appropriate references from the database in terms of both semantic and kinematic similarities. 2) Semantic-Modulated Transformer selectively absorbs retrieval knowledge, adapting to the difference between retrieved samples and the target motion sequence. 3) Condition Mixture better utilizes the retrieval database during inference, overcoming the scale sensitivity in classifier-free guidance. Extensive experiments demonstrate that ReMoDiffuse outperforms state-of-the-art methods by balancing both text-motion consistency and motion quality, especially for more diverse motion generation.

Jianglin Lu, Hailing Wang, Xu Ma et al.

Despite rapid progress, multimodal large language models (MLLMs) still fail on tasks that humans solve effortlessly, such as navigating a grid maze from a screenshot or selecting the correct puzzle piece. Rather than retraining the model, we ask a complementary question: how much capability can be elicited from a frozen MLLM purely by improving the execution scaffold around it? We introduce MUSE, a multimodal unified structured execution harness that wraps any off-the-shelf MLLM with composable modules for task representation, visual processing, perception tool use, structured parsing, deterministic verification, and verifier-guided repair, without any model retraining. We evaluate MUSE across diverse benchmarks spanning visual spatial planning, visual perception, multimodal reasoning, and fine-grained visual discrimination, using multiple state-of-the-art MLLMs. MUSE delivers consistent gains over the bare model in all settings, with the largest jumps on challenging instances. Further analysis reveals that many MLLM failures arise from harness-level shortcomings rather than fundamental model deficits, and can be addressed through verifier-guided repair without touching the model. These findings highlight the agentic multimodal harness as a critical yet underexplored design dimension, offering an orthogonal avenue for improving MLLMs beyond model-centric optimization.

Zhongang Cai, Liang Pan, Chen Wei et al.

Human pose and shape estimation (HPS) has attracted increasing attention in recent years. While most existing studies focus on HPS from 2D images or videos with inherent depth ambiguity, there are surging need to investigate HPS from 3D point clouds as depth sensors have been frequently employed in commercial devices. However, real-world sensory 3D points are usually noisy and incomplete, and also human bodies could have different poses of high diversity. To tackle these challenges, we propose a principled framework, PointHPS, for accurate 3D HPS from point clouds captured in real-world settings, which iteratively refines point features through a cascaded architecture. Specifically, each stage of PointHPS performs a series of downsampling and upsampling operations to extract and collate both local and global cues, which are further enhanced by two novel modules: 1) Cross-stage Feature Fusion (CFF) for multi-scale feature propagation that allows information to flow effectively through the stages, and 2) Intermediate Feature Enhancement (IFE) for body-aware feature aggregation that improves feature quality after each stage. To facilitate a comprehensive study under various scenarios, we conduct our experiments on two large-scale benchmarks, comprising i) a dataset that features diverse subjects and actions captured by real commercial sensors in a laboratory environment, and ii) controlled synthetic data generated with realistic considerations such as clothed humans in crowded outdoor scenes. Extensive experiments demonstrate that PointHPS, with its powerful point feature extraction and processing scheme, outperforms State-of-the-Art methods by significant margins across the board. Homepage: https://caizhongang.github.io/projects/PointHPS/.

Fangzhou Hong, Mingyuan Zhang, Liang Pan et al.

3D avatar creation plays a crucial role in the digital age. However, the whole production process is prohibitively time-consuming and labor-intensive. To democratize this technology to a larger audience, we propose AvatarCLIP, a zero-shot text-driven framework for 3D avatar generation and animation. Unlike professional software that requires expert knowledge, AvatarCLIP empowers layman users to customize a 3D avatar with the desired shape and texture, and drive the avatar with the described motions using solely natural languages. Our key insight is to take advantage of the powerful vision-language model CLIP for supervising neural human generation, in terms of 3D geometry, texture and animation. Specifically, driven by natural language descriptions, we initialize 3D human geometry generation with a shape VAE network. Based on the generated 3D human shapes, a volume rendering model is utilized to further facilitate geometry sculpting and texture generation. Moreover, by leveraging the priors learned in the motion VAE, a CLIP-guided reference-based motion synthesis method is proposed for the animation of the generated 3D avatar. Extensive qualitative and quantitative experiments validate the effectiveness and generalizability of AvatarCLIP on a wide range of avatars. Remarkably, AvatarCLIP can generate unseen 3D avatars with novel animations, achieving superior zero-shot capability.

Mingyuan Zhang, Yue Bai, Yifan Wang et al.

Explorations in fine-tuning Vision-Language Models (VLMs), such as Low-Rank Adaptation (LoRA) from Parameter Efficient Fine-Tuning (PEFT), have made impressive progress. However, most approaches rely on explicit weight updates, overlooking the extensive representational structures already encoded in pre-trained models that remain underutilized. Recent works have demonstrated that Mask Fine-Tuning (MFT) can be a powerful and efficient post-training paradigm for language models. Instead of updating weights, MFT assigns learnable gating scores to each weight, allowing the model to reorganize its internal subnetworks for downstream task adaptation. In this paper, we rethink fine-tuning for VLMs from a structural reparameterization perspective grounded in MFT. We apply MFT to the language and projector components of VLMs with different language backbones and compare against strong PEFT baselines. Experiments show that MFT consistently surpasses LoRA variants and even full fine-tuning, achieving high performance without altering the frozen backbone. Our findings reveal that effective adaptation can emerge not only from updating weights but also from reestablishing connections among the model's existing knowledge. Code available at: https://github.com/Ming-K9/MFT-VLM

Yiyang Huang, Yitian Zhang, Yizhou Wang et al.

Despite significant progress in video-language modeling, hallucinations remain a persistent challenge in Video Large Language Models (Vid-LLMs), referring to outputs that appear plausible yet contradict the content of the input video. This survey presents a comprehensive analysis of hallucinations in Vid-LLMs and introduces a systematic taxonomy that categorizes them into two core types: dynamic distortion and content fabrication, each comprising two subtypes with representative cases. Building on this taxonomy, we review recent advances in the evaluation and mitigation of hallucinations, covering key benchmarks, metrics, and intervention strategies. We further analyze the root causes of dynamic distortion and content fabrication, which often result from limited capacity for temporal representation and insufficient visual grounding. These insights inform several promising directions for future work, including the development of motion-aware visual encoders and the integration of counterfactual learning techniques. This survey consolidates scattered progress to foster a systematic understanding of hallucinations in Vid-LLMs, laying the groundwork for building robust and reliable video-language systems. An up-to-date curated list of related works is maintained at https://github.com/hukcc/Awesome-Video-Hallucination .

Mingyuan Zhang, Ambuj Tewari

In online ranking, a learning algorithm sequentially ranks a set of items and receives feedback on its ranking in the form of relevance scores. Since obtaining relevance scores typically involves human annotation, it is of great interest to consider a partial feedback setting where feedback is restricted to the top-$k$ items in the rankings. Chaudhuri and Tewari [2017] developed a framework to analyze online ranking algorithms with top $k$ feedback. A key element in their work was the use of techniques from partial monitoring. In this paper, we further investigate online ranking with top $k$ feedback and solve some open problems posed by Chaudhuri and Tewari [2017]. We provide a full characterization of minimax regret rates with the top $k$ feedback model for all $k$ and for the following ranking performance measures: Pairwise Loss, Discounted Cumulative Gain, and Precision@n. In addition, we give an efficient algorithm that achieves the minimax regret rate for Precision@n.

Chenyang Gu, Mingyuan Zhang, Haozhe Xie et al.

Prior motion generation largely follows two paradigms: continuous diffusion models that excel at kinematic control, and discrete token-based generators that are effective for semantic conditioning. To combine their strengths, we propose a three-stage framework comprising condition feature extraction (Perception), discrete token generation (Planning), and diffusion-based motion synthesis (Control). Central to this framework is MoTok, a diffusion-based discrete motion tokenizer that decouples semantic abstraction from fine-grained reconstruction by delegating motion recovery to a diffusion decoder, enabling compact single-layer tokens while preserving motion fidelity. For kinematic conditions, coarse constraints guide token generation during planning, while fine-grained constraints are enforced during control through diffusion-based optimization. This design prevents kinematic details from disrupting semantic token planning. On HumanML3D, our method significantly improves controllability and fidelity over MaskControl while using only one-sixth of the tokens, reducing trajectory error from 0.72 cm to 0.08 cm and FID from 0.083 to 0.029. Unlike prior methods that degrade under stronger kinematic constraints, ours improves fidelity, reducing FID from 0.033 to 0.014.

Wanqi Yin, Zhongang Cai, Ruisi Wang et al.

Expressive human pose and shape estimation (EHPS) unifies body, hands, and face motion capture with numerous applications. Despite encouraging progress, current state-of-the-art methods focus on training innovative architectural designs on confined datasets. In this work, we investigate the impact of scaling up EHPS towards a family of generalist foundation models. 1) For data scaling, we perform a systematic investigation on 40 EHPS datasets, encompassing a wide range of scenarios that a model trained on any single dataset cannot handle. More importantly, capitalizing on insights obtained from the extensive benchmarking process, we optimize our training scheme and select datasets that lead to a significant leap in EHPS capabilities. Ultimately, we achieve diminishing returns at 10M training instances from diverse data sources. 2) For model scaling, we take advantage of vision transformers (up to ViT-Huge as the backbone) to study the scaling law of model sizes in EHPS. To exclude the influence of algorithmic design, we base our experiments on two minimalist architectures: SMPLer-X, which consists of an intermediate step for hand and face localization, and SMPLest-X, an even simpler version that reduces the network to its bare essentials and highlights significant advances in the capture of articulated hands. With big data and the large model, the foundation models exhibit strong performance across diverse test benchmarks and excellent transferability to even unseen environments. Moreover, our finetuning strategy turns the generalist into specialist models, allowing them to achieve further performance boosts. Notably, our foundation models consistently deliver state-of-the-art results on seven benchmarks such as AGORA, UBody, EgoBody, and our proposed SynHand dataset for comprehensive hand evaluation. (Code is available at: https://github.com/wqyin/SMPLest-X).

Yukang Cao, Xinying Guo, Mingyuan Zhang et al.

While recent advances in text-to-motion generation have shown promising results, they typically assume all individuals are grouped as a single unit. Scaling these methods to handle larger crowds and ensuring that individuals respond appropriately to specific events remains a significant challenge. This is primarily due to the complexities of scene planning, which involves organizing groups, planning their activities, and coordinating interactions, and controllable motion generation. In this paper, we present CrowdMoGen, the first zero-shot framework for collective motion generation, which effectively groups individuals and generates event-aligned motion sequences from text prompts. 1) Being limited by the available datasets for training an effective scene planning module in a supervised manner, we instead propose a crowd scene planner that leverages pre-trained large language models (LLMs) to organize individuals into distinct groups. While LLMs offer high-level guidance for group divisions, they lack the low-level understanding of human motion. To address this, we further propose integrating an SMPL-based joint prior to generate context-appropriate activities, which consists of both joint trajectories and textual descriptions. 2) Secondly, to incorporate the assigned activities into the generative network, we introduce a collective motion generator that integrates the activities into a transformer-based network in a joint-wise manner, maintaining the spatial constraints during the multi-step denoising process. Extensive experiments demonstrate that CrowdMoGen significantly outperforms previous approaches, delivering realistic, event-driven motion sequences that are spatially coherent. As the first framework of collective motion generation, CrowdMoGen has the potential to advance applications in urban simulation, crowd planning, and other large-scale interactive environments.

Xingyu Zheng, Xianglong Liu, Haotong Qin et al.

With the advancement of diffusion models (DMs) and the substantially increased computational requirements, quantization emerges as a practical solution to obtain compact and efficient low-bit DMs. However, the highly discrete representation leads to severe accuracy degradation, hindering the quantization of diffusion models to ultra-low bit-widths. This paper proposes a novel weight binarization approach for DMs, namely BinaryDM, pushing binarized DMs to be accurate and efficient by improving the representation and optimization. From the representation perspective, we present an Evolvable-Basis Binarizer (EBB) to enable a smooth evolution of DMs from full-precision to accurately binarized. EBB enhances information representation in the initial stage through the flexible combination of multiple binary bases and applies regularization to evolve into efficient single-basis binarization. The evolution only occurs in the head and tail of the DM architecture to retain the stability of training. From the optimization perspective, a Low-rank Representation Mimicking (LRM) is applied to assist the optimization of binarized DMs. The LRM mimics the representations of full-precision DMs in low-rank space, alleviating the direction ambiguity of the optimization process caused by fine-grained alignment. Comprehensive experiments demonstrate that BinaryDM achieves significant accuracy and efficiency gains compared to SOTA quantization methods of DMs under ultra-low bit-widths. With 1-bit weight and 4-bit activation (W1A4), BinaryDM achieves as low as 7.74 FID and saves the performance from collapse (baseline FID 10.87). As the first binarization method for diffusion models, W1A4 BinaryDM achieves impressive 15.2x OPs and 29.2x model size savings, showcasing its substantial potential for edge deployment. The code is available at https://github.com/Xingyu-Zheng/BinaryDM.

Ronghui Li, Zhongyuan Hu, Li Siyao et al.

Although existing 3D dance generation methods perform well in controlled scenarios, they often struggle to generalize in the wild. When conditioned on unseen music, existing methods often produce unstructured or physically implausible dance, largely due to limited music-to-dance data and restricted model capacity. This work aims to push the frontier of generalizable 3D dance generation by scaling up both data and model design. (1) On the data side, we develop a fully automated pipeline that reconstructs high-fidelity 3D dance motions from monocular videos. To eliminate the physical artifacts prevalent in existing reconstruction methods, we introduce a Foot Restoration Diffusion Model (FRDM) guided by foot-contact and geometric constraints that enforce physical plausibility while preserving kinematic smoothness and expressiveness, resulting in a diverse, high-quality multimodal 3D dance dataset totaling 100.69 hours. (2) On model design, we propose Choreographic LLaMA (ChoreoLLaMA), a scalable LLaMA-based architecture. To enhance robustness under unfamiliar music conditions, we integrate a retrieval-augmented generation (RAG) module that injects reference dance as a prompt. Additionally, we design a slow/fast-cadence Mixture-of-Experts (MoE) module that enables ChoreoLLaMA to smoothly adapt motion rhythms across varying music tempos. Extensive experiments across diverse dance genres show that our approach surpasses existing methods in both qualitative and quantitative evaluations, marking a step toward scalable, real-world 3D dance generation. Code, models, and data will be released.

Zhikun Zhang, Yiting Duan, Xiangjun Wang et al.

This paper proposes a novel fast online methodology for outlier detection called the exception maximization outlier detection method(EMODM), which employs probabilistic models and statistical algorithms to detect abnormal patterns from the outputs of complex systems. The EMODM is based on a two-state Gaussian mixture model and demonstrates strong performance in probability anomaly detection working on real-time raw data rather than using special prior distribution information. We confirm this using the synthetic data from two numerical cases. For the real-world data, we have detected the short circuit pattern of the circuit system using EMODM by the current and voltage output of a three-phase inverter. The EMODM also found an abnormal period due to COVID-19 in the insured unemployment data of 53 regions in the United States from 2000 to 2024. The application of EMODM to these two real-life datasets demonstrated the effectiveness and accuracy of our algorithm.

Yizhou Wang, Lingzhi Zhang, Yue Bai et al.

Next token prediction paradigm has been prevailing for autoregressive models in the era of LLMs. The current default sampling choice for popular LLMs is temperature scaling together with nucleus sampling to balance diversity and coherence. Nevertheless, such approach leads to inferior performance in various NLP tasks when the model is not certain about testing questions. To this end, we propose a brand new training-free decoding strategy, dubbed as Cautious Next Token Prediction (CNTP). In the decoding process, if the model has comparatively high prediction entropy at a certain step, we sample multiple trials starting from the step independently and stop when encountering any punctuation. Then we select the trial with the lowest perplexity score viewed as the most probable and reliable trial path given the model's capacity. The trial number is negatively correlated with the prediction confidence, i.e., the less confident the model is, the more trials it should sample. This is consistent with human beings' behaviour: when feeling uncertain or unconfident, one tends to think more creatively, exploring multiple thinking paths, to cautiously select the path one feels most confident about. Extensive experiments on both LLMs and MLLMs show that our proposed CNTP approach outperforms existing standard decoding strategies consistently by a clear margin. Moreover, the integration of CNTP with self consistency can further improve over vanilla self consistency. We believe our proposed CNTP has the potential to become one of the default choices for LLM decoding. Code is available at https://github.com/wyzjack/CNTP.

Mingyuan Zhang, Zhicheng Zhang, Hao Wu et al.

We present flow matching for reaction coordinates (FMRC), a novel deep learning algorithm designed to identify optimal reaction coordinates (RC) in biomolecular reversible dynamics. FMRC is based on the mathematical principles of lumpability and decomposability, which we reformulate into a conditional probability framework for efficient data-driven optimization using deep generative models. While FMRC does not explicitly learn the well-established transfer operator or its eigenfunctions, it can effectively encode the dynamics of leading eigenfunctions of the system transfer operator into its low-dimensional RC space. We further quantitatively compare its performance with several state-of-the-art algorithms by evaluating the quality of Markov state models (MSM) constructed in their respective RC spaces, demonstrating the superiority of FMRC in three increasingly complex biomolecular systems. In addition, we successfully demonstrated the efficacy of FMRC for bias deposition in the enhanced sampling of a simple model system. Finally, we discuss its potential applications in downstream applications such as enhanced sampling methods and MSM construction.

Zhuoran Shen, Mingyuan Zhang, Haiyu Zhao et al.

Dot-product attention has wide applications in computer vision and natural language processing. However, its memory and computational costs grow quadratically with the input size. Such growth prohibits its application on high-resolution inputs. To remedy this drawback, this paper proposes a novel efficient attention mechanism equivalent to dot-product attention but with substantially less memory and computational costs. Its resource efficiency allows more widespread and flexible integration of attention modules into a network, which leads to better accuracies. Empirical evaluations demonstrated the effectiveness of its advantages. Efficient attention modules brought significant performance boosts to object detectors and instance segmenters on MS-COCO 2017. Further, the resource efficiency democratizes attention to complex models, where high costs prohibit the use of dot-product attention. As an exemplar, a model with efficient attention achieved state-of-the-art accuracies for stereo depth estimation on the Scene Flow dataset. Code is available at https://github.com/cmsflash/efficient-attention.

Mingyuan Zhang, Huirong Li, Zhongang Cai et al.

Text-driven motion generation has achieved substantial progress with the emergence of diffusion models. However, existing methods still struggle to generate complex motion sequences that correspond to fine-grained descriptions, depicting detailed and accurate spatio-temporal actions. This lack of fine controllability limits the usage of motion generation to a larger audience. To tackle these challenges, we present FineMoGen, a diffusion-based motion generation and editing framework that can synthesize fine-grained motions, with spatial-temporal composition to the user instructions. Specifically, FineMoGen builds upon diffusion model with a novel transformer architecture dubbed Spatio-Temporal Mixture Attention (SAMI). SAMI optimizes the generation of the global attention template from two perspectives: 1) explicitly modeling the constraints of spatio-temporal composition; and 2) utilizing sparsely-activated mixture-of-experts to adaptively extract fine-grained features. To facilitate a large-scale study on this new fine-grained motion generation task, we contribute the HuMMan-MoGen dataset, which consists of 2,968 videos and 102,336 fine-grained spatio-temporal descriptions. Extensive experiments validate that FineMoGen exhibits superior motion generation quality over state-of-the-art methods. Notably, FineMoGen further enables zero-shot motion editing capabilities with the aid of modern large language models (LLM), which faithfully manipulates motion sequences with fine-grained instructions. Project Page: https://mingyuan-zhang.github.io/projects/FineMoGen.html

Mingyuan Zhang, Daisheng Jin, Chenyang Gu et al.

Human motion generation, a cornerstone technique in animation and video production, has widespread applications in various tasks like text-to-motion and music-to-dance. Previous works focus on developing specialist models tailored for each task without scalability. In this work, we present Large Motion Model (LMM), a motion-centric, multi-modal framework that unifies mainstream motion generation tasks into a generalist model. A unified motion model is appealing since it can leverage a wide range of motion data to achieve broad generalization beyond a single task. However, it is also challenging due to the heterogeneous nature of substantially different motion data and tasks. LMM tackles these challenges from three principled aspects: 1) Data: We consolidate datasets with different modalities, formats and tasks into a comprehensive yet unified motion generation dataset, MotionVerse, comprising 10 tasks, 16 datasets, a total of 320k sequences, and 100 million frames. 2) Architecture: We design an articulated attention mechanism ArtAttention that incorporates body part-aware modeling into Diffusion Transformer backbone. 3) Pre-Training: We propose a novel pre-training strategy for LMM, which employs variable frame rates and masking forms, to better exploit knowledge from diverse training data. Extensive experiments demonstrate that our generalist LMM achieves competitive performance across various standard motion generation tasks over state-of-the-art specialist models. Notably, LMM exhibits strong generalization capabilities and emerging properties across many unseen tasks. Additionally, our ablation studies reveal valuable insights about training and scaling up large motion models for future research.

Zhongang Cai, Jianping Jiang, Zhongfei Qing et al.

In this work, we present Digital Life Project, a framework utilizing language as the universal medium to build autonomous 3D characters, who are capable of engaging in social interactions and expressing with articulated body motions, thereby simulating life in a digital environment. Our framework comprises two primary components: 1) SocioMind: a meticulously crafted digital brain that models personalities with systematic few-shot exemplars, incorporates a reflection process based on psychology principles, and emulates autonomy by initiating dialogue topics; 2) MoMat-MoGen: a text-driven motion synthesis paradigm for controlling the character's digital body. It integrates motion matching, a proven industry technique to ensure motion quality, with cutting-edge advancements in motion generation for diversity. Extensive experiments demonstrate that each module achieves state-of-the-art performance in its respective domain. Collectively, they enable virtual characters to initiate and sustain dialogues autonomously, while evolving their socio-psychological states. Concurrently, these characters can perform contextually relevant bodily movements. Additionally, a motion captioning module further allows the virtual character to recognize and appropriately respond to human players' actions. Homepage: https://digital-life-project.com/

Kehong Gong, Zhengyu Wen, Dao Thien Phong et al.

Recent methods for arbitrary-skeleton motion capture from monocular video follow a factorized pipeline, where a Video-to-Pose network predicts joint positions and an analytical inverse-kinematics (IK) stage recovers joint rotations. While effective, this design is inherently limited, since joint positions do not fully determine rotations and leave degrees of freedom such as bone-axis twist ambiguous, and the non-differentiable IK stage prevents the system from adapting to noisy predictions or optimizing for the final animation objective. In this work, we present the first fully end-to-end framework in which both Video-to-Pose and Pose-to-Rotation are learnable and jointly optimized. We observe that the ambiguity in pose-to-rotation mapping arises from missing coordinate system information: the same joint positions can correspond to different rotations under different rest poses and local axis conventions. To resolve this, we introduce a reference pose-rotation pair from the target asset, which, together with the rest pose, not only anchors the mapping but also defines the underlying rotation coordinate system. This formulation turns rotation prediction into a well-constrained conditional problem and enables effective learning. In addition, our model predicts joint positions directly from video without relying on mesh intermediates, improving both robustness and efficiency. Both stages share a skeleton-aware Global-Local Graph-guided Multi-Head Attention (GL-GMHA) module for joint-level local reasoning and global coordination. Experiments on Truebones Zoo and Objaverse show that our method reduces rotation error from ~17 degrees to ~10 degrees, and to 6.54 degrees on unseen skeletons, while achieving ~20x faster inference than mesh-based pipelines. Project page: https://animotionlab.github.io/MoCapAnythingV2/

Jiawei Ren, Mingyuan Zhang, Cunjun Yu et al.

Generating animation of physics-based characters with intuitive control has long been a desirable task with numerous applications. However, generating physically simulated animations that reflect high-level human instructions remains a difficult problem due to the complexity of physical environments and the richness of human language. In this paper, we present InsActor, a principled generative framework that leverages recent advancements in diffusion-based human motion models to produce instruction-driven animations of physics-based characters. Our framework empowers InsActor to capture complex relationships between high-level human instructions and character motions by employing diffusion policies for flexibly conditioned motion planning. To overcome invalid states and infeasible state transitions in planned motions, InsActor discovers low-level skills and maps plans to latent skill sequences in a compact latent space. Extensive experiments demonstrate that InsActor achieves state-of-the-art results on various tasks, including instruction-driven motion generation and instruction-driven waypoint heading. Notably, the ability of InsActor to generate physically simulated animations using high-level human instructions makes it a valuable tool, particularly in executing long-horizon tasks with a rich set of instructions.

Kehong Gong, Zhengyu Wen, Weixia He et al.

Motion capture now underpins content creation far beyond digital humans, yet most existing pipelines remain species- or template-specific. We formalize this gap as Category-Agnostic Motion Capture (CAMoCap): given a monocular video and an arbitrary rigged 3D asset as a prompt, the goal is to reconstruct a rotation-based animation such as BVH that directly drives the specific asset. We present MoCapAnything, a reference-guided, factorized framework that first predicts 3D joint trajectories and then recovers asset-specific rotations via constraint-aware inverse kinematics. The system contains three learnable modules and a lightweight IK stage: (1) a Reference Prompt Encoder that extracts per-joint queries from the asset's skeleton, mesh, and rendered images; (2) a Video Feature Extractor that computes dense visual descriptors and reconstructs a coarse 4D deforming mesh to bridge the gap between video and joint space; and (3) a Unified Motion Decoder that fuses these cues to produce temporally coherent trajectories. We also curate Truebones Zoo with 1038 motion clips, each providing a standardized skeleton-mesh-render triad. Experiments on both in-domain benchmarks and in-the-wild videos show that MoCapAnything delivers high-quality skeletal animations and exhibits meaningful cross-species retargeting across heterogeneous rigs, enabling scalable, prompt-driven 3D motion capture for arbitrary assets. Project page: https://animotionlab.github.io/MoCapAnything/

Kehong Gong, Zhengyu Wen, Mingxi Xu et al.

Despite significant progress in 4D content generation, the conversion of monocular videos into high-quality animated 3D assets with explicit 4D meshes remains considerably challenging. The scarcity of large-scale, naturally captured 4D mesh datasets further limits the ability to train generalizable video-to-4D models from scratch in a purely data-driven manner. Meanwhile, advances in image-to-3D generation, supported by extensive datasets, offer powerful prior models that can be leveraged. To better utilize these priors while minimizing reliance on 4D supervision, we introduce SWiT-4D, a Sliding-Window Transformer for lossless, parameter-free temporal 4D mesh generation. SWiT-4D integrates seamlessly with any Diffusion Transformer (DiT)-based image-to-3D generator, adding spatial-temporal modeling across video frames while preserving the original single-image forward process, enabling 4D mesh reconstruction from videos of arbitrary length. To recover global translation, we further introduce an optimization-based trajectory module tailored for static-camera monocular videos. SWiT-4D demonstrates strong data efficiency: with only a single short (<10s) video for fine-tuning, it achieves high-fidelity geometry and stable temporal consistency, indicating practical deployability under extremely limited 4D supervision. Comprehensive experiments on both in-domain zoo-test sets and challenging out-of-domain benchmarks (C4D, Objaverse, and in-the-wild videos) show that SWiT-4D consistently outperforms existing baselines in temporal smoothness. Project page: https://animotionlab.github.io/SWIT4D/

Mingxi Xu, Qi Wang, Zhengyu Wen et al.

Motion tokenization is a key component of generalizable motion models, yet most existing approaches are restricted to species-specific skeletons, limiting their applicability across diverse morphologies. We propose NECromancer (NEC), a universal motion tokenizer that operates directly on arbitrary BVH skeletons. NEC consists of three components: (1) an Ontology-aware Skeletal Graph Encoder (OwO) that encodes structural priors from BVH files, including joint semantics, rest-pose offsets, and skeletal topology, into skeletal embeddings; (2) a Topology-Agnostic Tokenizer (TAT) that compresses motion sequences into a universal, topology-invariant discrete representation; and (3) the Unified BVH Universe (UvU), a large-scale dataset aggregating BVH motions across heterogeneous skeletons. Experiments show that NEC achieves high-fidelity reconstruction under substantial compression and effectively disentangles motion from skeletal structure. The resulting token space supports cross-species motion transfer, composition, denoising, generation with token-based models, and text-motion retrieval, establishing a unified framework for motion analysis and synthesis across diverse morphologies. Demo page: https://animotionlab.github.io/NECromancer/

Ning Zhang, Zhengyu Li, Kwong Weng Loh et al.

Prior masked modeling motion generation methods predominantly study text-to-motion. We present DiMo, a discrete diffusion-style framework, which extends masked modeling to bidirectional text--motion understanding and generation. Unlike GPT-style autoregressive approaches that tokenize motion and decode sequentially, DiMo performs iterative masked token refinement, unifying Text-to-Motion (T2M), Motion-to-Text (M2T), and text-free Motion-to-Motion (M2M) within a single model. This decoding paradigm naturally enables a quality-latency trade-off at inference via the number of refinement steps.We further improve motion token fidelity with residual vector quantization (RVQ) and enhance alignment and controllability with Group Relative Policy Optimization (GRPO). Experiments on HumanML3D and KIT-ML show strong motion quality and competitive bidirectional understanding under a unified framework. In addition, we demonstrate model ability in text-free motion completion, text-guided motion prediction and motion caption correction without architectural change.Additional qualitative results are available on our project page: https://animotionlab.github.io/DiMo/.

Yi Xu, Ruining Yang, Yitian Zhang et al.

Recent advances in large language models (LLMs) have sparked growing interest in integrating language-driven techniques into trajectory prediction. By leveraging their semantic and reasoning capabilities, LLMs are reshaping how autonomous systems perceive, model, and predict trajectories. This survey provides a comprehensive overview of this emerging field, categorizing recent work into five directions: (1) Trajectory prediction via language modeling paradigms, (2) Direct trajectory prediction with pretrained language models, (3) Language-guided scene understanding for trajectory prediction, (4) Language-driven data generation for trajectory prediction, (5) Language-based reasoning and interpretability for trajectory prediction. For each, we analyze representative methods, highlight core design choices, and identify open challenges. This survey bridges natural language processing and trajectory prediction, offering a unified perspective on how language can enrich trajectory prediction.

Zhouyingcheng Liao, Mingyuan Zhang, Wenjia Wang et al.

While motion generation has made substantial progress, its practical application remains constrained by dataset diversity and scale, limiting its ability to handle out-of-distribution scenarios. To address this, we propose a simple and effective baseline, RMD, which enhances the generalization of motion generation through retrieval-augmented techniques. Unlike previous retrieval-based methods, RMD requires no additional training and offers three key advantages: (1) the external retrieval database can be flexibly replaced; (2) body parts from the motion database can be reused, with an LLM facilitating splitting and recombination; and (3) a pre-trained motion diffusion model serves as a prior to improve the quality of motions obtained through retrieval and direct combination. Without any training, RMD achieves state-of-the-art performance, with notable advantages on out-of-distribution data.

Ryuma Nakahata, Shehtab Zaman, Mingyuan Zhang et al.

Ptychography is a computational method of microscopy that recovers high-resolution transmission images of samples from a series of diffraction patterns. While conventional phase retrieval algorithms can iteratively recover the images, they require oversampled diffraction patterns, incur significant computational costs, and struggle to recover the absolute phase of the sample's transmission function. Deep learning algorithms for ptychography are a promising approach to resolving the limitations of iterative algorithms. We present PtychoFormer, a hierarchical transformer-based model for data-driven single-shot ptychographic phase retrieval. PtychoFormer processes subsets of diffraction patterns, generating local inferences that are seamlessly stitched together to produce a high-quality reconstruction. Our model exhibits tolerance to sparsely scanned diffraction patterns and achieves up to 3600 times faster imaging speed than the extended ptychographic iterative engine (ePIE). We also propose the extended-PtychoFormer (ePF), a hybrid approach that combines the benefits of PtychoFormer with the ePIE. ePF minimizes global phase shifts and significantly enhances reconstruction quality, achieving state-of-the-art phase retrieval in ptychography.

Mingyuan Zhang, Shivani Agarwal

There has been much interest in recent years in learning good classifiers from data with noisy labels. Most work on learning from noisy labels has focused on standard loss-based performance measures. However, many machine learning problems require using non-decomposable performance measures which cannot be expressed as the expectation or sum of a loss on individual examples; these include for example the H-mean, Q-mean and G-mean in class imbalance settings, and the Micro $F_1$ in information retrieval. In this paper, we design algorithms to learn from noisy labels for two broad classes of multiclass non-decomposable performance measures, namely, monotonic convex and ratio-of-linear, which encompass all the above examples. Our work builds on the Frank-Wolfe and Bisection based methods of Narasimhan et al. (2015). In both cases, we develop noise-corrected versions of the algorithms under the widely studied family of class-conditional noise models. We provide regret (excess risk) bounds for our algorithms, establishing that even though they are trained on noisy data, they are Bayes consistent in the sense that their performance converges to the optimal performance w.r.t. the clean (non-noisy) distribution. Our experiments demonstrate the effectiveness of our algorithms in handling label noise.

Zi-An Wang, Shihao Zou, Shiyao Yu et al.

Recent advances in interactive technologies have highlighted the prominence of audio signals for semantic encoding. This paper explores a new task, where audio signals are used as conditioning inputs to generate motions that align with the semantics of the audio. Unlike text-based interactions, audio provides a more natural and intuitive communication method. However, existing methods typically focus on matching motions with music or speech rhythms, which often results in a weak connection between the semantics of the audio and generated motions. We propose an end-to-end framework using a masked generative transformer, enhanced by a memory-retrieval attention module to handle sparse and lengthy audio inputs. Additionally, we enrich existing datasets by converting descriptions into conversational style and generating corresponding audio with varied speaker identities. Experiments demonstrate the effectiveness and efficiency of the proposed framework, demonstrating that audio instructions can convey semantics similar to text while providing more practical and user-friendly interactions.

Mingyuan Zhang, Yue Bai, Huan Wang et al.

The model is usually kept integral in the mainstream large language model (LLM) fine-tuning protocols. No works have questioned whether maintaining the integrity of the model is indispensable for performance. In this work, we introduce Mask Fine-Tuning (MFT), a brand-new LLM fine-tuning paradigm to show that properly breaking the integrity of the model can surprisingly lead to improved performance. Specifically, MFT learns a set of binary masks supervised by the typical LLM fine-tuning objective. Extensive experiments show that MFT gains a consistent performance boost across various domains and backbones (e.g., 1.95%/1.88% average gain in coding with LLaMA2-7B/3.1-8B). Detailed procedures are provided to study the proposed MFT from different hyperparameter perspectives for better insight. In particular, MFT naturally updates the current LLM training protocol by deploying it on a complete well-trained model. This study extends the functionality of mask learning from its conventional network pruning context for model compression to a more general scope.

Lei Yang, Junshan Jin, Mingyuan Zhang et al.

Visual speech recognition is a technique to identify spoken content in silent speech videos, which has raised significant attention in recent years. Advancements in data-driven deep learning methods have significantly improved both the speed and accuracy of recognition. However, these deep learning methods can be effected by visual disturbances, such as lightning conditions, skin texture and other user-specific features. Data-driven approaches could reduce the performance degradation caused by these visual disturbances using models pretrained on large-scale datasets. But these methods often require large amounts of training data and computational resources, making them costly. To reduce the influence of user-specific features and enhance performance with limited data, this paper proposed a landmark guided visual feature extractor. Facial landmarks are used as auxiliary information to aid in training the visual feature extractor. A spatio-temporal multi-graph convolutional network is designed to fully exploit the spatial locations and spatio-temporal features of facial landmarks. Additionally, a multi-level lip dynamic fusion framework is introduced to combine the spatio-temporal features of the landmarks with the visual features extracted from the raw video frames. Experimental results show that this approach performs well with limited data and also improves the model's accuracy on unseen speakers.

Shiyao Yu, Zi-An Wang, Kangning Yin et al.

Motion retrieval is crucial for motion acquisition, offering superior precision, realism, controllability, and editability compared to motion generation. Existing approaches leverage contrastive learning to construct a unified embedding space for motion retrieval from text or visual modality. However, these methods lack a more intuitive and user-friendly interaction mode and often overlook the sequential representation of most modalities for improved retrieval performance. To address these limitations, we propose a framework that aligns four modalities -- text, audio, video, and motion -- within a fine-grained joint embedding space, incorporating audio for the first time in motion retrieval to enhance user immersion and convenience. This fine-grained space is achieved through a sequence-level contrastive learning approach, which captures critical details across modalities for better alignment. To evaluate our framework, we augment existing text-motion datasets with synthetic but diverse audio recordings, creating two multi-modal motion retrieval datasets. Experimental results demonstrate superior performance over state-of-the-art methods across multiple sub-tasks, including an 10.16% improvement in R@10 for text-to-motion retrieval and a 25.43% improvement in R@1 for video-to-motion retrieval on the HumanML3D dataset. Furthermore, our results show that our 4-modal framework significantly outperforms its 3-modal counterpart, underscoring the potential of multi-modal motion retrieval for advancing motion acquisition.

Chongzhi Zhang, Mingyuan Zhang, Zhiyang Teng et al.

Natural Language Video Localization (NLVL), grounding phrases from natural language descriptions to corresponding video segments, is a complex yet critical task in video understanding. Despite ongoing advancements, many existing solutions lack the capability to globally capture temporal dynamics of the video data. In this study, we present a novel approach to NLVL that aims to address this issue. Our method involves the direct generation of a global 2D temporal map via a conditional denoising diffusion process, based on the input video and language query. The main challenges are the inherent sparsity and discontinuity of a 2D temporal map in devising the diffusion decoder. To address these challenges, we introduce a multi-scale technique and develop an innovative diffusion decoder. Our approach effectively encapsulates the interaction between the query and video data across various time scales. Experiments on the Charades and DiDeMo datasets underscore the potency of our design.

Jiawei Ren, Mingyuan Zhang, Cunjun Yu et al.

Data imbalance exists ubiquitously in real-world visual regressions, e.g., age estimation and pose estimation, hurting the model's generalizability and fairness. Thus, imbalanced regression gains increasing research attention recently. Compared to imbalanced classification, imbalanced regression focuses on continuous labels, which can be boundless and high-dimensional and hence more challenging. In this work, we identify that the widely used Mean Square Error (MSE) loss function can be ineffective in imbalanced regression. We revisit MSE from a statistical view and propose a novel loss function, Balanced MSE, to accommodate the imbalanced training label distribution. We further design multiple implementations of Balanced MSE to tackle different real-world scenarios, particularly including the one that requires no prior knowledge about the training label distribution. Moreover, to the best of our knowledge, Balanced MSE is the first general solution to high-dimensional imbalanced regression. Extensive experiments on both synthetic and three real-world benchmarks demonstrate the effectiveness of Balanced MSE.

Zhongang Cai, Mingyuan Zhang, Jiawei Ren et al.

Image- and video-based 3D human recovery (i.e., pose and shape estimation) have achieved substantial progress. However, due to the prohibitive cost of motion capture, existing datasets are often limited in scale and diversity. In this work, we obtain massive human sequences by playing the video game with automatically annotated 3D ground truths. Specifically, we contribute GTA-Human, a large-scale 3D human dataset generated with the GTA-V game engine, featuring a highly diverse set of subjects, actions, and scenarios. More importantly, we study the use of game-playing data and obtain five major insights. First, game-playing data is surprisingly effective. A simple frame-based baseline trained on GTA-Human outperforms more sophisticated methods by a large margin. For video-based methods, GTA-Human is even on par with the in-domain training set. Second, we discover that synthetic data provides critical complements to the real data that is typically collected indoor. Our investigation into domain gap provides explanations for our data mixture strategies that are simple yet useful. Third, the scale of the dataset matters. The performance boost is closely related to the additional data available. A systematic study reveals the model sensitivity to data density from multiple key aspects. Fourth, the effectiveness of GTA-Human is also attributed to the rich collection of strong supervision labels (SMPL parameters), which are otherwise expensive to acquire in real datasets. Fifth, the benefits of synthetic data extend to larger models such as deeper convolutional neural networks (CNNs) and Transformers, for which a significant impact is also observed. We hope our work could pave the way for scaling up 3D human recovery to the real world. Homepage: https://caizhongang.github.io/projects/GTA-Human/

Ricardo Bigolin Lanfredi, Mingyuan Zhang, William F. Auffermann et al.

Deep learning has shown recent success in classifying anomalies in chest x-rays, but datasets are still small compared to natural image datasets. Supervision of abnormality localization has been shown to improve trained models, partially compensating for dataset sizes. However, explicitly labeling these anomalies requires an expert and is very time-consuming. We propose a potentially scalable method for collecting implicit localization data using an eye tracker to capture gaze locations and a microphone to capture a dictation of a report, imitating the setup of a reading room. The resulting REFLACX (Reports and Eye-Tracking Data for Localization of Abnormalities in Chest X-rays) dataset was labeled across five radiologists and contains 3,032 synchronized sets of eye-tracking data and timestamped report transcriptions for 2,616 chest x-rays from the MIMIC-CXR dataset. We also provide auxiliary annotations, including bounding boxes around lungs and heart and validation labels consisting of ellipses localizing abnormalities and image-level labels. Furthermore, a small subset of the data contains readings from all radiologists, allowing for the calculation of inter-rater scores.

Daxuan Ren, Jianmin Zheng, Jianfei Cai et al.

Generating an interpretable and compact representation of 3D shapes from point clouds is an important and challenging problem. This paper presents CSG-Stump Net, an unsupervised end-to-end network for learning shapes from point clouds and discovering the underlying constituent modeling primitives and operations as well. At the core is a three-level structure called {\em CSG-Stump}, consisting of a complement layer at the bottom, an intersection layer in the middle, and a union layer at the top. CSG-Stump is proven to be equivalent to CSG in terms of representation, therefore inheriting the interpretable, compact and editable nature of CSG while freeing from CSG's complex tree structures. Particularly, the CSG-Stump has a simple and regular structure, allowing neural networks to give outputs of a constant dimensionality, which makes itself deep-learning friendly. Due to these characteristics of CSG-Stump, CSG-Stump Net achieves superior results compared to previous CSG-based methods and generates much more appealing shapes, as confirmed by extensive experiments. Project page: https://kimren227.github.io/projects/CSGStump/

Chongzhi Zhang, Mingyuan Zhang, Shanghang Zhang et al.

Vision Transformers (ViTs) have achieved impressive performance on various vision tasks, yet their generalization under distribution shifts (DS) is rarely understood. In this work, we comprehensively study the out-of-distribution (OOD) generalization of ViTs. For systematic investigation, we first present a taxonomy of DS. We then perform extensive evaluations of ViT variants under different DS and compare their generalization with Convolutional Neural Network (CNN) models. Important observations are obtained: 1) ViTs learn weaker biases on backgrounds and textures, while they are equipped with stronger inductive biases towards shapes and structures, which is more consistent with human cognitive traits. Therefore, ViTs generalize better than CNNs under DS. With the same or less amount of parameters, ViTs are ahead of corresponding CNNs by more than 5% in top-1 accuracy under most types of DS. 2) As the model scale increases, ViTs strengthen these biases and thus gradually narrow the in-distribution and OOD performance gap. To further improve the generalization of ViTs, we design the Generalization-Enhanced ViTs (GE-ViTs) from the perspectives of adversarial learning, information theory, and self-supervised learning. By comprehensively investigating these GE-ViTs and comparing with their corresponding CNN models, we observe: 1) For the enhanced model, larger ViTs still benefit more for the OOD generalization. 2) GE-ViTs are more sensitive to the hyper-parameters than their corresponding CNN models. We design a smoother learning strategy to achieve a stable training process and obtain performance improvements on OOD data by 4% from vanilla ViTs. We hope our comprehensive study could shed light on the design of more generalizable learning architectures.

Daisheng Jin, Xiao Ma, Chongzhi Zhang et al.

In this paper, we investigate the cause of the high false positive rate in Visual Relationship Detection (VRD). We observe that during training, the relationship proposal distribution is highly imbalanced: most of the negative relationship proposals are easy to identify, e.g., the inaccurate object detection, which leads to the under-fitting of low-frequency difficult proposals. This paper presents Spatially-Aware Balanced negative pRoposal sAmpling (SABRA), a robust VRD framework that alleviates the influence of false positives. To effectively optimize the model under imbalanced distribution, SABRA adopts Balanced Negative Proposal Sampling (BNPS) strategy for mini-batch sampling. BNPS divides proposals into 5 well defined sub-classes and generates a balanced training distribution according to the inverse frequency. BNPS gives an easier optimization landscape and significantly reduces the number of false positives. To further resolve the low-frequency challenging false positive proposals with high spatial ambiguity, we improve the spatial modeling ability of SABRA on two aspects: a simple and efficient multi-head heterogeneous graph attention network (MH-GAT) that models the global spatial interactions of objects, and a spatial mask decoder that learns the local spatial configuration. SABRA outperforms SOTA methods by a large margin on two human-object interaction (HOI) datasets and one general VRD dataset.

Haotong Qin, Zhongang Cai, Mingyuan Zhang et al.

To alleviate the resource constraint for real-time point cloud applications that run on edge devices, in this paper we present BiPointNet, the first model binarization approach for efficient deep learning on point clouds. We discover that the immense performance drop of binarized models for point clouds mainly stems from two challenges: aggregation-induced feature homogenization that leads to a degradation of information entropy, and scale distortion that hinders optimization and invalidates scale-sensitive structures. With theoretical justifications and in-depth analysis, our BiPointNet introduces Entropy-Maximizing Aggregation (EMA) to modulate the distribution before aggregation for the maximum information entropy, and Layer-wise Scale Recovery (LSR) to efficiently restore feature representation capacity. Extensive experiments show that BiPointNet outperforms existing binarization methods by convincing margins, at the level even comparable with the full precision counterpart. We highlight that our techniques are generic, guaranteeing significant improvements on various fundamental tasks and mainstream backbones. Moreover, BiPointNet gives an impressive 14.7x speedup and 18.9x storage saving on real-world resource-constrained devices.

Mingyuan Zhang, Harish G. Ramaswamy, Shivani Agarwal

The F-measure is a widely used performance measure for multi-label classification, where multiple labels can be active in an instance simultaneously (e.g. in image tagging, multiple tags can be active in any image). In particular, the F-measure explicitly balances recall (fraction of active labels predicted to be active) and precision (fraction of labels predicted to be active that are actually so), both of which are important in evaluating the overall performance of a multi-label classifier. As with most discrete prediction problems, however, directly optimizing the F-measure is computationally hard. In this paper, we explore the question of designing convex surrogate losses that are calibrated for the F-measure -- specifically, that have the property that minimizing the surrogate loss yields (in the limit of sufficient data) a Bayes optimal multi-label classifier for the F-measure. We show that the F-measure for an $s$-label problem, when viewed as a $2^s \times 2^s$ loss matrix, has rank at most $s^2+1$, and apply a result of Ramaswamy et al. (2014) to design a family of convex calibrated surrogates for the F-measure. The resulting surrogate risk minimization algorithms can be viewed as decomposing the multi-label F-measure learning problem into $s^2+1$ binary class probability estimation problems. We also provide a quantitative regret transfer bound for our surrogates, which allows any regret guarantees for the binary problems to be transferred to regret guarantees for the overall F-measure problem, and discuss a connection with the algorithm of Dembczynski et al. (2013). Our experiments confirm our theoretical findings.

Hangfeng He, Mingyuan Zhang, Qiang Ning et al.

Real-world applications often require improved models by leveraging a range of cheap incidental supervision signals. These could include partial labels, noisy labels, knowledge-based constraints, and cross-domain or cross-task annotations -- all having statistical associations with gold annotations but not exactly the same. However, we currently lack a principled way to measure the benefits of these signals to a given target task, and the common practice of evaluating these benefits is through exhaustive experiments with various models and hyperparameters. This paper studies whether we can, in a single framework, quantify the benefits of various types of incidental signals for a given target task without going through combinatorial experiments. We propose a unified PAC-Bayesian motivated informativeness measure, PABI, that characterizes the uncertainty reduction provided by incidental supervision signals. We demonstrate PABI's effectiveness by quantifying the value added by various types of incidental signals to sequence tagging tasks. Experiments on named entity recognition (NER) and question answering (QA) show that PABI's predictions correlate well with learning performance, providing a promising way to determine, ahead of learning, which supervision signals would be beneficial.