Hongjun Wang

Yitong Jiang, Hongjun Wang, Collin McCarthy et al.

Vision foundation models are bottlenecked by the quadratic cost of self-attention, which limits usable resolution and increases the cost of large-scale pretraining. Subquadratic alternatives such as linear attention and state-space models reduce this cost, but often serialize images into 1D token streams and weaken the 2D spatial structure important for vision. Generalized Spatial Propagation Networks (GSPN) instead propagate context directly on the 2D grid through line-scan recurrences, achieving near-linear complexity without positional embeddings, but have seen little use as foundation-scale encoders. We present C-GSPN, a foundation-scale vision encoder based on 2D spatial propagation. C-GSPN makes the operator practical through three improvements: (1) a fast GSPN CUDA kernel that fuses per-step launches into a single warp-specialized implementation with shared-memory tiling, coalesced access, and a compact multi-channel propagation, reaching over 90% of peak memory bandwidth and running up to 40--52x faster than the original GSPN implementation; (2) a compressed latent-space propagation block with fused normalization, which turns kernel-level speed into block- and model-level efficiency; and (3) a two-stage cross-operator distillation recipe that trains the new architecture from an attention teacher without the cost of from-scratch foundation-scale training. Distilled with 600M image-text pairs, C-GSPN matches an isomorphic ViT baseline with 15% fewer parameters, improves ADE20K segmentation by +2.1%, transfers to high resolution with a fraction of the data needed from scratch, and delivers a 4x end-to-end block speedup at 2K with single-pass, tiling-free inference.

Hongjun Wang, Sagar Vaze, Kai Han

Detecting test-time distribution shift has emerged as a key capability for safely deployed machine learning models, with the question being tackled under various guises in recent years. In this paper, we aim to provide a consolidated view of the two largest sub-fields within the community: out-of-distribution (OOD) detection and open-set recognition (OSR). In particular, we aim to provide rigorous empirical analysis of different methods across settings and provide actionable takeaways for practitioners and researchers. Concretely, we make the following contributions: (i) We perform rigorous cross-evaluation between state-of-the-art methods in the OOD detection and OSR settings and identify a strong correlation between the performances of methods for them; (ii) We propose a new, large-scale benchmark setting which we suggest better disentangles the problem tackled by OOD detection and OSR, re-evaluating state-of-the-art OOD detection and OSR methods in this setting; (iii) We surprisingly find that the best performing method on standard benchmarks (Outlier Exposure) struggles when tested at scale, while scoring rules which are sensitive to the deep feature magnitude consistently show promise; and (iv) We conduct empirical analysis to explain these phenomena and highlight directions for future research. Code: https://github.com/Visual-AI/Dissect-OOD-OSR

Hongjun Wang, Yisen Wang

Deep neural networks obtained by standard training have been constantly plagued by adversarial examples. Although adversarial training demonstrates its capability to defend against adversarial examples, unfortunately, it leads to an inevitable drop in the natural generalization. To address the issue, we decouple the natural generalization and the robust generalization from joint training and formulate different training strategies for each one. Specifically, instead of minimizing a global loss on the expectation over these two generalization errors, we propose a bi-expert framework called \emph{Generalist} where we simultaneously train base learners with task-aware strategies so that they can specialize in their own fields. The parameters of base learners are collected and combined to form a global learner at intervals during the training process. The global learner is then distributed to the base learners as initialized parameters for continued training. Theoretically, we prove that the risks of Generalist will get lower once the base learners are well trained. Extensive experiments verify the applicability of Generalist to achieve high accuracy on natural examples while maintaining considerable robustness to adversarial ones. Code is available at https://github.com/PKU-ML/Generalist.

Zhiwen Zhang, Hongjun Wang, Jiyuan Chen et al.

Estimating the travel time of a path is an essential topic for intelligent transportation systems. It serves as the foundation for real-world applications, such as traffic monitoring, route planning, and taxi dispatching. However, building a model for such a data-driven task requires a large amount of users' travel information, which directly relates to their privacy and thus is less likely to be shared. The non-Independent and Identically Distributed (non-IID) trajectory data across data owners also make a predictive model extremely challenging to be personalized if we directly apply federated learning. Finally, previous work on travel time estimation does not consider the real-time traffic state of roads, which we argue can significantly influence the prediction. To address the above challenges, we introduce GOF-TTE for the mobile user group, Generative Online Federated Learning Framework for Travel Time Estimation, which I) utilizes the federated learning approach, allowing private data to be kept on client devices while training, and designs the global model as an online generative model shared by all clients to infer the real-time road traffic state. II) apart from sharing a base model at the server, adapts a fine-tuned personalized model for every client to study their personal driving habits, making up for the residual error made by localized global model prediction. % III) designs the global model as an online generative model shared by all clients to infer the real-time road traffic state. We also employ a simple privacy attack to our framework and implement the differential privacy mechanism to further guarantee privacy safety. Finally, we conduct experiments on two real-world public taxi datasets of DiDi Chengdu and Xi'an. The experimental results demonstrate the effectiveness of our proposed framework.

Hongjun Wang, Yisen Wang

Due to numerous breakthroughs in real-world applications brought by machine intelligence, deep neural networks (DNNs) are widely employed in critical applications. However, predictions of DNNs are easily manipulated with imperceptible adversarial perturbations, which impedes the further deployment of DNNs and may result in profound security and privacy implications. By incorporating adversarial samples into the training data pool, adversarial training is the strongest principled strategy against various adversarial attacks among all sorts of defense methods. Recent works mainly focus on developing new loss functions or regularizers, attempting to find the unique optimal point in the weight space. But none of them taps the potentials of classifiers obtained from standard adversarial training, especially states on the searching trajectory of training. In this work, we are dedicated to the weight states of models through the training process and devise a simple but powerful \emph{Self-Ensemble Adversarial Training} (SEAT) method for yielding a robust classifier by averaging weights of history models. This considerably improves the robustness of the target model against several well known adversarial attacks, even merely utilizing the naive cross-entropy loss to supervise. We also discuss the relationship between the ensemble of predictions from different adversarially trained models and the prediction of weight-ensembled models, as well as provide theoretical and empirical evidence that the proposed self-ensemble method provides a smoother loss landscape and better robustness than both individual models and the ensemble of predictions from different classifiers. We further analyze a subtle but fatal issue in the general settings for the self-ensemble model, which causes the deterioration of the weight-ensembled method in the late phases.

Hongjun Wang, Zhiwen Zhang, Zipei Fan et al.

Travel time estimation from GPS trips is of great importance to order duration, ridesharing, taxi dispatching, etc. However, the dense trajectory is not always available due to the limitation of data privacy and acquisition, while the origin destination (OD) type of data, such as NYC taxi data, NYC bike data, and Capital Bikeshare data, is more accessible. To address this issue, this paper starts to estimate the OD trips travel time combined with the road network. Subsequently, a Multitask Weakly Supervised Learning Framework for Travel Time Estimation (MWSL TTE) has been proposed to infer transition probability between roads segments, and the travel time on road segments and intersection simultaneously. Technically, given an OD pair, the transition probability intends to recover the most possible route. And then, the output of travel time is equal to the summation of all segments' and intersections' travel time in this route. A novel route recovery function has been proposed to iteratively maximize the current route's co occurrence probability, and minimize the discrepancy between routes' probability distribution and the inverse distribution of routes' estimation loss. Moreover, the expected log likelihood function based on a weakly supervised framework has been deployed in optimizing the travel time from road segments and intersections concurrently. We conduct experiments on a wide range of real world taxi datasets in Xi'an and Chengdu and demonstrate our method's effectiveness on route recovery and travel time estimation.

Hongjun Wang, Sagar Vaze, Kai Han

Generalized Category Discovery (GCD) is a challenging task in which, given a partially labelled dataset, models must categorize all unlabelled instances, regardless of whether they come from labelled categories or from new ones. In this paper, we challenge a remaining assumption in this task: that all images share the same domain. Specifically, we introduce a new task and method to handle GCD when the unlabelled data also contains images from different domains to the labelled set. Our proposed `HiLo' networks extract High-level semantic and Low-level domain features, before minimizing the mutual information between the representations. Our intuition is that the clusterings based on domain information and semantic information should be independent. We further extend our method with a specialized domain augmentation tailored for the GCD task, as well as a curriculum learning approach. Finally, we construct a benchmark from corrupted fine-grained datasets as well as a large-scale evaluation on DomainNet with real-world domain shifts, reimplementing a number of GCD baselines in this setting. We demonstrate that HiLo outperforms SoTA category discovery models by a large margin on all evaluations.

Hongjun Wang, Jiyuan Chen, Tong Pan et al.

Spatial-temporal (ST) graph modeling, such as traffic speed forecasting and taxi demand prediction, is an important task in deep learning area. However, for the nodes in graph, their ST patterns can vary greatly in difficulties for modeling, owning to the heterogeneous nature of ST data. We argue that unveiling the nodes to the model in a meaningful order, from easy to complex, can provide performance improvements over traditional training procedure. The idea has its root in Curriculum Learning which suggests in the early stage of training models can be sensitive to noise and difficult samples. In this paper, we propose ST-Curriculum Dropout, a novel and easy-to-implement strategy for spatial-temporal graph modeling. Specifically, we evaluate the learning difficulty of each node in high-level feature space and drop those difficult ones out to ensure the model only needs to handle fundamental ST relations at the beginning, before gradually moving to hard ones. Our strategy can be applied to any canonical deep learning architecture without extra trainable parameters, and extensive experiments on a wide range of datasets are conducted to illustrate that, by controlling the difficulty level of ST relations as the training progresses, the model is able to capture better representation of the data and thus yields better generalization.

Zhiwen Zhang, Hongjun Wang, Zipei Fan et al.

Due to the rapid development of Internet of Things (IoT) technologies, many online web apps (e.g., Google Map and Uber) estimate the travel time of trajectory data collected by mobile devices. However, in reality, complex factors, such as network communication and energy constraints, make multiple trajectories collected at a low sampling rate. In this case, this paper aims to resolve the problem of travel time estimation (TTE) and route recovery in sparse scenarios, which often leads to the uncertain label of travel time and route between continuously sampled GPS points. We formulate this problem as an inexact supervision problem in which the training data has coarsely grained labels and jointly solve the tasks of TTE and route recovery. And we argue that both two tasks are complementary to each other in the model-learning procedure and hold such a relation: more precise travel time can lead to better inference for routes, in turn, resulting in a more accurate time estimation). Based on this assumption, we propose an EM algorithm to alternatively estimate the travel time of inferred route through weak supervision in E step and retrieve the route based on estimated travel time in M step for sparse trajectories. We conducted experiments on three real-world trajectory datasets and demonstrated the effectiveness of the proposed method.

Hongjun Wang, Jiyuan Chen, Zipei Fan et al.

Recently, forecasting the crowd flows has become an important research topic, and plentiful technologies have achieved good performances. As we all know, the flow at a citywide level is in a mixed state with several basic patterns (e.g., commuting, working, and commercial) caused by the city area functional distributions (e.g., developed commercial areas, educational areas and parks). However, existing technologies have been criticized for their lack of considering the differences in the flow patterns among regions since they want to build only one comprehensive model to learn the mixed flow tensors. Recognizing this limitation, we present a new perspective on flow prediction and propose an explainable framework named ST-ExpertNet, which can adopt every spatial-temporal model and train a set of functional experts devoted to specific flow patterns. Technically, we train a bunch of experts based on the Mixture of Experts (MoE), which guides each expert to specialize in different kinds of flow patterns in sample spaces by using the gating network. We define several criteria, including comprehensiveness, sparsity, and preciseness, to construct the experts for better interpretability and performances. We conduct experiments on a wide range of real-world taxi and bike datasets in Beijing and NYC. The visualizations of the expert's intermediate results demonstrate that our ST-ExpertNet successfully disentangles the city's mixed flow tensors along with the city layout, e.g., the urban ring road structure. Different network architectures, such as ST-ResNet, ConvLSTM, and CNN, have been adopted into our ST-ExpertNet framework for experiments and the results demonstrates the superiority of our framework in both interpretability and performances.

Jiawei Wang, Chuang Yang, Jiawei Yong et al.

Mobility trajectories are essential for understanding urban dynamics and enhancing urban planning, yet access to such data is frequently hindered by privacy concerns. This research introduces a transformative framework for generating large-scale urban mobility trajectories, employing a novel application of a transformer-based model pre-trained and fine-tuned through a two-phase process. Initially, trajectory generation is conceptualized as an offline reinforcement learning (RL) problem, with a significant reduction in vocabulary space achieved during tokenization. The integration of Inverse Reinforcement Learning (IRL) allows for the capture of trajectory-wise reward signals, leveraging historical data to infer individual mobility preferences. Subsequently, the pre-trained model is fine-tuned using the constructed reward model, effectively addressing the challenges inherent in traditional RL-based autoregressive methods, such as long-term credit assignment and handling of sparse reward environments. Comprehensive evaluations on multiple datasets illustrate that our framework markedly surpasses existing models in terms of reliability and diversity. Our findings not only advance the field of urban mobility modeling but also provide a robust methodology for simulating urban data, with significant implications for traffic management and urban development planning. The implementation is publicly available at https://github.com/Wangjw6/TrajGPT_R.

Yifan Jiang, Zezheng Feng, Hongjun Wang et al.

In recent years, deep learning approaches have been proved good performance in traffic flow prediction, many complex models have been proposed to make traffic flow prediction more accurate. However, lacking transparency limits the domain experts on understanding when and where the input data mainly impact the results. Most urban experts and planners can only adjust traffic based on their own experience and can not react effectively toward the potential traffic jam. To tackle this problem, we adapt Shapley value and present a visualization analysis system , which can provide experts with the interpretation of traffic flow prediction. TrafPS consists of three layers, from data process to results computation and visualization. We design three visualization views in TrafPS to support the prediction analysis process. One demonstration shows that the TrafPS supports an effective analytical pipeline on interpreting the prediction flow to users and provides an intuitive visualization for decision making.

Inclusion AI, Tiwei Bie, Haoxing Chen et al.

We present LLaDA2.0-Uni, a unified discrete diffusion large language model (dLLM) that supports multimodal understanding and generation within a natively integrated framework. Its architecture combines a fully semantic discrete tokenizer, a MoE-based dLLM backbone, and a diffusion decoder. By discretizing continuous visual inputs via SigLIP-VQ, the model enables block-level masked diffusion for both text and vision inputs within the backbone, while the decoder reconstructs visual tokens into high-fidelity images. Inference efficiency is enhanced beyond parallel decoding through prefix-aware optimizations in the backbone and few-step distillation in the decoder. Supported by carefully curated large-scale data and a tailored multi-stage training pipeline, LLaDA2.0-Uni matches specialized VLMs in multimodal understanding while delivering strong performance in image generation and editing. Its native support for interleaved generation and reasoning establishes a promising and scalable paradigm for next-generation unified foundation models. Codes and models are available at https://github.com/inclusionAI/LLaDA2.0-Uni.

Yuxuan He, Chaiming Huang, Yifan Wu et al.

A short video succeeds not simply because of what it shows, but because of how it schedules attention -- yet current multimodal models lack the structural grammar to parse or produce this organization. Existing models can describe scenes, answer event-centric questions, and read on-screen text, but they are far less reliable at identifying timeline-grounded units such as hooks, cut rationales, shot-induced tension, and platform-facing packaging cues. We propose SV6D (Structured Video in Six Dimensions), inspired by professional storyboard practice in film and television production, a representation framework that decomposes internet-native video into six complementary structural dimensions -- subject, aesthetics, camera language, editing, narrative, and dissemination -- with each label tied to physically observable evidence on the timeline. We formalize a unified optimization objective over SV6D that combines Hungarian-matched temporal alignment, dimension-wise semantic label distance, and quality regularization. Building on this framework, we present Leum-VL-8B, an 8B video-language model that realizes the SV6D objective through an expert-driven post-training pipeline, further refined through verifiable reinforcement learning on perception-oriented tasks. Leum-VL-8B achieves 70.8 on VideoMME (w/o subtitles), 70.0 on MVBench, and 61.6 on MotionBench, while remaining competitive on general multimodal evaluations such as MMBench-EN. We also construct FeedBench, a benchmark for structure-sensitive short-video understanding. Our results indicate that the missing layer in video AI is not pixel generation but structural representation: grounded on the timeline, linked to visible evidence, and directly consumable by downstream workflows such as editing, retrieval, recommendation, and generation control, including text-heavy internet video formats with overlays and image-text layouts.

Zetian Shen, Hongjun Wang, Jiyuan Chen et al.

Supply chains are integral to global economic stability, yet disruptions can swiftly propagate through interconnected networks, resulting in substantial economic impacts. Accurate and timely inference of supply chain resilience the capability to maintain core functions during disruptions is crucial for proactive risk mitigation and robust network design. However, existing approaches lack effective mechanisms to infer supply chain resilience without explicit system dynamics and struggle to represent the higher-order, multi-entity dependencies inherent in supply chain networks. These limitations motivate the definition of a novel problem and the development of targeted modeling solutions. To address these challenges, we formalize a novel problem: Supply Chain Resilience Inference (SCRI), defined as predicting supply chain resilience using hypergraph topology and observed inventory trajectories without explicit dynamic equations. To solve this problem, we propose the Supply Chain Resilience Inference Hypergraph Network (SC-RIHN), a novel hypergraph-based model leveraging set-based encoding and hypergraph message passing to capture multi-party firm-product interactions. Comprehensive experiments demonstrate that SC-RIHN significantly outperforms traditional MLP, representative graph neural network variants, and ResInf baselines across synthetic benchmarks, underscoring its potential for practical, early-warning risk assessment in complex supply chain systems.

Hongjun Wang, Jiyuan Chen, Xuan Song et al.

Post-training quantization offers an efficient pathway to deploy super-resolution models, yet existing methods treat weight and activation quantization independently, missing their critical interplay. Through controlled experiments on SwinIR, we uncover a striking asymmetry: weight quantization primarily degrades structural similarity, while activation quantization disproportionately affects pixel-level accuracy. This stems from their distinct roles--weights encode learned restoration priors for textures and edges, whereas activations carry input-specific intensity information. Building on this insight, we propose HarmoQ, a unified framework that harmonizes quantization across components through three synergistic steps: structural residual calibration proactively adjusts weights to compensate for activation-induced detail loss, harmonized scale optimization analytically balances quantization difficulty via closed-form solutions, and adaptive boundary refinement iteratively maintains this balance during optimization. Experiments show HarmoQ achieves substantial gains under aggressive compression, outperforming prior art by 0.46 dB on Set5 at 2-bit while delivering 3.2x speedup and 4x memory reduction on A100 GPUs. This work provides the first systematic analysis of weight-activation coupling in super-resolution quantization and establishes a principled solution for efficient high-quality image restoration.

Hongjun Wang, Wei Liu, Weibo Gu et al.

Regulating the importance ratio is critical for the training stability of Group Relative Policy Optimization (GRPO) based frameworks. However, prevailing ratio control methods, such as hard clipping, suffer from non-differentiable boundaries and vanishing gradient regions, failing to maintain gradient fidelity. Furthermore, these methods lack a hazard-aware mechanism to adaptively suppress extreme deviations, leaving the optimization process vulnerable to abrupt policy shifts. To address these challenges, we propose Modulated Hazard-aware Policy Optimization (MHPO), a novel framework designed for robust and stable reinforcement learning. The proposed MHPO introduces a Log-Fidelity Modulator (LFM) to map unbounded importance ratios into a bounded, differentiable domain. This mechanism effectively prevents high-variance outlier tokens from destabilizing the loss landscape while ensuring global gradient stability. Complementarily, a Decoupled Hazard Penalty (DHP) integrates cumulative hazard functions from survival analysis to independently regulate positive and negative policy shifts. By shaping the optimization landscape with hazard-aware penalties, the proposed MHPO achieves fine-grained regulation of asymmetric policy shifts simultaneously mitigating mode collapse from over-expansion and preventing policy erosion from catastrophic contraction within a stabilized trust region. Extensive evaluations on diverse reasoning benchmarks across both text-based and vision-language tasks demonstrate that MHPO consistently outperforms existing methods, achieving superior performance while significantly enhancing training stability.

Hongjun Wang, Jiawei Yong, Jiawei Wang et al.

Traffic prediction remains a key challenge in spatio-temporal data mining, despite progress in deep learning. Accurate forecasting is hindered by the complex influence of external factors such as traffic accidents and regulations, often overlooked by existing models due to limited data integration. To address these limitations, we present two enriched traffic datasets from Tokyo and California, incorporating traffic accident and regulation data. Leveraging these datasets, we propose ConFormer (Conditional Transformer), a novel framework that integrates graph propagation with guided normalization layer. This design dynamically adjusts spatial and temporal node relationships based on historical patterns, enhancing predictive accuracy. Our model surpasses the state-of-the-art STAEFormer in both predictive performance and efficiency, achieving lower computational costs and reduced parameter demands. Extensive evaluations demonstrate that ConFormer consistently outperforms mainstream spatio-temporal baselines across multiple metrics, underscoring its potential to advance traffic prediction research.

Mingsong Li, Lin Liu, Hongjun Wang et al.

Current instruction-based image editing (IBIE) methods struggle with challenging editing tasks, as both editing types and sample counts of existing datasets are limited. Moreover, traditional dataset construction often contains noisy image-caption pairs, which may introduce biases and limit model capabilities in complex editing scenarios. To address these limitations, we introduce MultiEdit, a comprehensive dataset featuring over 107K high-quality image editing samples. It encompasses 6 challenging editing tasks through a diverse collection of 18 non-style-transfer editing types and 38 style transfer operations, covering a spectrum from sophisticated style transfer to complex semantic operations like person reference editing and in-image text editing. We employ a novel dataset construction pipeline that utilizes two multi-modal large language models (MLLMs) to generate visual-adaptive editing instructions and produce high-fidelity edited images, respectively. Extensive experiments demonstrate that fine-tuning foundational open-source models with our MultiEdit-Train set substantially improves models' performance on sophisticated editing tasks in our proposed MultiEdit-Test benchmark, while effectively preserving their capabilities on the standard editing benchmark. We believe MultiEdit provides a valuable resource for advancing research into more diverse and challenging IBIE capabilities. Our dataset is available at https://huggingface.co/datasets/inclusionAI/MultiEdit.

Kun-Yu Lin, Hongjun Wang, Weining Ren et al.

This work introduces panoptic captioning, a novel task striving to seek the minimum text equivalence of images. We take the first step towards panoptic captioning by formulating it as a task of generating a comprehensive textual description for an image, which encapsulates all entities, their respective locations and attributes, relationships among entities, as well as global image state. Through an extensive evaluation, our work reveals that state-of-the-art Multi-modal Large Language Models (MLLMs) have limited performance in solving panoptic captioning. To address this, we propose an effective data engine named PancapEngine to produce high-quality data and a novel method named PancapChain to improve panoptic captioning. Specifically, our PancapEngine first detects diverse categories of entities in images by an elaborate detection suite, and then generates required panoptic captions using entity-aware prompts. Additionally, our PancapChain explicitly decouples the challenging panoptic captioning task into multiple stages and generates panoptic captions step by step. More importantly, we contribute a comprehensive metric named PancapScore and a human-curated test set for reliable model evaluation. Experiments show that our PancapChain-13B model can beat state-of-the-art open-source MLLMs like InternVL-2.5-78B and even surpass proprietary models like GPT-4o and Gemini-2.0-Pro, demonstrating the effectiveness of our data engine and method. Project page: https://visual-ai.github.io/pancap/

Hongjun Wang, Guangrun Wang, Ya Li et al.

The success of DNNs has driven the extensive applications of person re-identification (ReID) into a new era. However, whether ReID inherits the vulnerability of DNNs remains unexplored. To examine the robustness of ReID systems is rather important because the insecurity of ReID systems may cause severe losses, e.g., the criminals may use the adversarial perturbations to cheat the CCTV systems. In this work, we examine the insecurity of current best-performing ReID models by proposing a learning-to-mis-rank formulation to perturb the ranking of the system output. As the cross-dataset transferability is crucial in the ReID domain, we also perform a back-box attack by developing a novel multi-stage network architecture that pyramids the features of different levels to extract general and transferable features for the adversarial perturbations. Our method can control the number of malicious pixels by using differentiable multi-shot sampling. To guarantee the inconspicuousness of the attack, we also propose a new perception loss to achieve better visual quality. Extensive experiments on four of the largest ReID benchmarks (i.e., Market1501 [45], CUHK03 [18], DukeMTMC [33], and MSMT17 [40]) not only show the effectiveness of our method, but also provides directions of the future improvement in the robustness of ReID systems. For example, the accuracy of one of the best-performing ReID systems drops sharply from 91.8% to 1.4% after being attacked by our method. Some attack results are shown in Fig. 1. The code is available at https://github.com/whj363636/Adversarial-attack-on-Person-ReID-With-Deep-Mis-Ranking.

Hongjun Wang, Sagar Vaze, Kai Han

Generalized Category Discovery (GCD) aims to classify unlabelled images from both `seen' and `unseen' classes by transferring knowledge from a set of labelled `seen' class images. A key theme in existing GCD approaches is adapting large-scale pre-trained models for the GCD task. An alternate perspective, however, is to adapt the data representation itself for better alignment with the pre-trained model. As such, in this paper, we introduce a two-stage adaptation approach termed SPTNet, which iteratively optimizes model parameters (i.e., model-finetuning) and data parameters (i.e., prompt learning). Furthermore, we propose a novel spatial prompt tuning method (SPT) which considers the spatial property of image data, enabling the method to better focus on object parts, which can transfer between seen and unseen classes. We thoroughly evaluate our SPTNet on standard benchmarks and demonstrate that our method outperforms existing GCD methods. Notably, we find our method achieves an average accuracy of 61.4% on the SSB, surpassing prior state-of-the-art methods by approximately 10%. The improvement is particularly remarkable as our method yields extra parameters amounting to only 0.117% of those in the backbone architecture. Project page: https://visual-ai.github.io/sptnet.

Xin Li, Yeying Jin, Xin Jin et al.

This paper reviews the NTIRE 2025 Challenge on Day and Night Raindrop Removal for Dual-Focused Images. This challenge received a wide range of impressive solutions, which are developed and evaluated using our collected real-world Raindrop Clarity dataset. Unlike existing deraining datasets, our Raindrop Clarity dataset is more diverse and challenging in degradation types and contents, which includes day raindrop-focused, day background-focused, night raindrop-focused, and night background-focused degradations. This dataset is divided into three subsets for competition: 14,139 images for training, 240 images for validation, and 731 images for testing. The primary objective of this challenge is to establish a new and powerful benchmark for the task of removing raindrops under varying lighting and focus conditions. There are a total of 361 participants in the competition, and 32 teams submitting valid solutions and fact sheets for the final testing phase. These submissions achieved state-of-the-art (SOTA) performance on the Raindrop Clarity dataset. The project can be found at https://lixinustc.github.io/CVPR-NTIRE2025-RainDrop-Competition.github.io/.

Hongjun Wang, Po Hu, Kai Han

Generalized Category Discovery (GCD) aims to categorize unlabelled instances from both known and unknown classes by transferring knowledge from labelled data of known classes. Existing methods assume all data comes from a single domain, yet real-world unlabelled data often exhibits domain shifts alongside semantic shifts. We study GCD under domain shifts and propose three frameworks that adapt foundation models, ranging from self-supervised vision models to vision-language models. (i) HiLo disentangles domain and semantic features through multi-level feature extraction and mutual information minimization, combined with PatchMix augmentation and curriculum sampling. (ii) HLPrompt extends HiLo with semantic-aware spatial prompt tuning to suppress background and domain noise. (iii) VLPrompt leverages vision-language models via factorized textual prompts and cross-modal consistency regularization. The three methods share core design principles while operating on different foundation backbones, making them suitable for different deployment scenarios. Extensive experiments on synthetic corruptions and real-world multi-domain shifts demonstrate consistent improvements over strong baselines. Project page: https://visual-ai.github.io/hilo/

Hongjun Wang, Jiyuan Chen, Yinqiang Zheng et al.

Deep learning has led to a dramatic leap on Single Image Super-Resolution (SISR) performances in recent years. %Despite the substantial advancement% While most existing work assumes a simple and fixed degradation model (e.g., bicubic downsampling), the research of Blind SR seeks to improve model generalization ability with unknown degradation. Recently, Kong et al pioneer the investigation of a more suitable training strategy for Blind SR using Dropout. Although such method indeed brings substantial generalization improvements via mitigating overfitting, we argue that Dropout simultaneously introduces undesirable side-effect that compromises model's capacity to faithfully reconstruct fine details. We show both the theoretical and experimental analyses in our paper, and furthermore, we present another easy yet effective training strategy that enhances the generalization ability of the model by simply modulating its first and second-order features statistics. Experimental results have shown that our method could serve as a model-agnostic regularization and outperforms Dropout on seven benchmark datasets including both synthetic and real-world scenarios.

Hongjun Wang, Wonmin Byeon, Jiarui Xu et al.

We present the Generalized Spatial Propagation Network (GSPN), a new attention mechanism optimized for vision tasks that inherently captures 2D spatial structures. Existing attention models, including transformers, linear attention, and state-space models like Mamba, process multi-dimensional data as 1D sequences, compromising spatial coherence and efficiency. GSPN overcomes these limitations by directly operating on spatially coherent image data and forming dense pairwise connections through a line-scan approach. Central to GSPN is the Stability-Context Condition, which ensures stable, context-aware propagation across 2D sequences and reduces the effective sequence length to $\sqrt{N}$ for a square map with N elements, significantly enhancing computational efficiency. With learnable, input-dependent weights and no reliance on positional embeddings, GSPN achieves superior spatial fidelity and state-of-the-art performance in vision tasks, including ImageNet classification, class-guided image generation, and text-to-image generation. Notably, GSPN accelerates SD-XL with softmax-attention by over $84\times$ when generating 16K images.

Zezheng Feng, Yifan Jiang, Hongjun Wang et al.

Recent achievements in deep learning (DL) have shown its potential for predicting traffic flows. Such predictions are beneficial for understanding the situation and making decisions in traffic control. However, most state-of-the-art DL models are considered "black boxes" with little to no transparency for end users with respect to the underlying mechanisms. Some previous work tried to "open the black boxes" and increase the interpretability of how predictions are generated. However, it still remains challenging to handle complex models on large-scale spatio-temporal data and discover salient spatial and temporal patterns that significantly influence traffic flows. To overcome the challenges, we present TrafPS, a visual analytics approach for interpreting traffic prediction outcomes to support decision-making in traffic management and urban planning. The measurements, region SHAP and trajectory SHAP, are proposed to quantify the impact of flow patterns on urban traffic at different levels. Based on the task requirement from the domain experts, we employ an interactive visual interface for multi-aspect exploration and analysis of significant flow patterns. Two real-world case studies demonstrate the effectiveness of TrafPS in identifying key routes and decision-making support for urban planning.

Yisen Wang, Yichuan Mo, Hongjun Wang et al.

Despite the rapid progress of neural networks, they remain highly vulnerable to adversarial examples, for which adversarial training (AT) is currently the most effective defense. While AT has been extensively studied, its practical applications expose two major limitations: natural accuracy tends to degrade significantly compared with standard training, and robustness does not transfer well across attacks crafted under different norm constraints. Unlike prior works that attempt to address only one issue within a single network, we propose to partition the overall generalization goal into multiple sub-tasks, each assigned to a dedicated base learner. By specializing in its designated objective, each base learner quickly becomes an expert in its field. In the later stages of training, we interpolate their parameters to form a knowledgeable global learner, while periodically redistributing the global parameters back to the base learners to prevent their optimization trajectories from drifting too far from the shared target. We term this framework Generalist and introduce three variants tailored to different application scenarios. Both theoretical analysis and extensive experiments demonstrate that Generalist achieves lower generalization error and significantly alleviates the trade-off problems compared with baseline methods. Our results suggest that Generalist provides a promising step toward developing fully robust classifiers in the future.

Hongjun Wang, Jiyuan Chen, Lingyu Zhang et al.

Spatiotemporal Graph Neural Networks (ST-GNNs) and Transformers have shown significant promise in traffic forecasting by effectively modeling temporal and spatial correlations. However, rapid urbanization in recent years has led to dynamic shifts in traffic patterns and travel demand, posing major challenges for accurate long-term traffic prediction. The generalization capability of ST-GNNs in extended temporal scenarios and cross-city applications remains largely unexplored. In this study, we evaluate state-of-the-art models on an extended traffic benchmark and observe substantial performance degradation in existing ST-GNNs over time, which we attribute to their limited inductive capabilities. Our analysis reveals that this degradation stems from an inability to adapt to evolving spatial relationships within urban environments. To address this limitation, we reconsider the design of adaptive embeddings and propose a Principal Component Analysis (PCA) embedding approach that enables models to adapt to new scenarios without retraining. We incorporate PCA embeddings into existing ST-GNN and Transformer architectures, achieving marked improvements in performance. Notably, PCA embeddings allow for flexibility in graph structures between training and testing, enabling models trained on one city to perform zero-shot predictions on other cities. This adaptability demonstrates the potential of PCA embeddings in enhancing the robustness and generalization of spatiotemporal models.

Gengming Zhang, Hao Cao, Kewei Hu et al.

Accurately identifying lychee-picking points in unstructured orchard environments and obtaining their coordinate locations is critical to the success of lychee-picking robots. However, traditional two-dimensional (2D) image-based object detection methods often struggle due to the complex geometric structures of branches, leaves and fruits, leading to incorrect determination of lychee picking points. In this study, we propose a Fcaf3d-lychee network model specifically designed for the accurate localisation of lychee picking points. Point cloud data of lychee picking points in natural environments are acquired using Microsoft's Azure Kinect DK time-of-flight (TOF) camera through multi-view stitching. We augment the Fully Convolutional Anchor-Free 3D Object Detection (Fcaf3d) model with a squeeze-and-excitation(SE) module, which exploits human visual attention mechanisms for improved feature extraction of lychee picking points. The trained network model is evaluated on a test set of lychee-picking locations and achieves an impressive F1 score of 88.57%, significantly outperforming existing models. Subsequent three-dimensional (3D) position detection of picking points in real lychee orchard environments yields high accuracy, even under varying degrees of occlusion. Localisation errors of lychee picking points are within 1.5 cm in all directions, demonstrating the robustness and generality of the model.

Hongjun Wang, Yitong Jiang, Collin McCarthy et al.

Efficient vision transformer remains a bottleneck for high-resolution images and long-video related real-world applications. Generalized Spatial Propagation Network (GSPN) addresses this by replacing quadratic self-attention with a line-scan propagation scheme, bringing the cost close to linear in the number of rows or columns, while retaining accuracy. Despite this advancement, the existing GSPN implementation still suffers from (i) heavy overhead due to repeatedly launching GPU kernels, (ii) excessive data transfers from global GPU memory, and (iii) redundant computations caused by maintaining separate propagation weights for each channel. We introduce GSPN-2, a joint algorithm-system redesign. In particular, we eliminate thousands of micro-launches from the previous implementation into one single 2D kernel, explicitly pin one warp to each channel slice, and stage the previous column's activations in shared memory. On the model side, we introduce a compact channel propagation strategy that replaces per-channel matrices, trimming parameters, and align naturally with the affinity map used in transformer attention. Experiments demonstrate GSPN-2's effectiveness across image classification and text-to-image synthesis tasks, matching transformer-level accuracy with significantly lower computational cost. GSPN-2 establishes a new efficiency frontier for modeling global spatial context in vision applications through its unique combination of structured matrix transformations and GPU-optimized implementation. Project page: https://whj363636.github.io/GSPN2/

Weining Ren, Hongjun Wang, Xiao Tan et al.

We present Fin3R, a simple, effective, and general fine-tuning method for feed-forward 3D reconstruction models. The family of feed-forward reconstruction model regresses pointmap of all input images to a reference frame coordinate system, along with other auxiliary outputs, in a single forward pass. However, we find that current models struggle with fine geometry and robustness due to (\textit{i}) the scarcity of high-fidelity depth and pose supervision and (\textit{ii}) the inherent geometric misalignment from multi-view pointmap regression. Fin3R jointly tackles two issues with an extra lightweight fine-tuning step. We freeze the decoder, which handles view matching, and fine-tune only the image encoder-the component dedicated to feature extraction. The encoder is enriched with fine geometric details distilled from a strong monocular teacher model on large, unlabeled datasets, using a custom, lightweight LoRA adapter. We validate our method on a wide range of models, including DUSt3R, MASt3R, CUT3R, and VGGT. The fine-tuned models consistently deliver sharper boundaries, recover complex structures, and achieve higher geometric accuracy in both single- and multi-view settings, while adding only the tiny LoRA weights, which leave test-time memory and latency virtually unchanged. Project page: \href{http://visual-ai.github.io/fin3r}{https://visual-ai.github.io/fin3r}

Chen Liang, Zhaoqi Huang, Haofen Wang et al.

Large language models (LLMs), as a novel information technology, are seeing increasing adoption in the Architecture, Engineering, and Construction (AEC) field. They have shown their potential to streamline processes throughout the building lifecycle. However, the robustness and reliability of LLMs in such a specialized and safety-critical domain remain to be evaluated. To address this challenge, this paper establishes AECBench, a comprehensive benchmark designed to quantify the strengths and limitations of current LLMs in the AEC domain. The benchmark defines 23 representative tasks within a five-level cognition-oriented evaluation framework encompassing Knowledge Memorization, Understanding, Reasoning, Calculation, and Application. These tasks were derived from authentic AEC practice, with scope ranging from codes retrieval to specialized documents generation. Subsequently, a 4,800-question dataset encompassing diverse formats, including open-ended questions, was crafted primarily by engineers and validated through a two-round expert review. Furthermore, an LLM-as-a-Judge approach was introduced to provide a scalable and consistent methodology for evaluating complex, long-form responses leveraging expert-derived rubrics. Through the evaluation of nine LLMs, a clear performance decline across five cognitive levels was revealed. Despite demonstrating proficiency in foundational tasks at the Knowledge Memorization and Understanding levels, the models showed significant performance deficits, particularly in interpreting knowledge from tables in building codes, executing complex reasoning and calculation, and generating domain-specific documents. Consequently, this study lays the groundwork for future research and development aimed at the robust and reliable integration of LLMs into safety-critical engineering practices.

Hongjun Wang, Jiyuan Chen, Zhengwei Yin et al.

Generalizable Image Super-Resolution aims to enhance model generalization capabilities under unknown degradations. To achieve this goal, the models are expected to focus only on image content-related features instead of overfitting degradations. Recently, numerous approaches such as Dropout and Feature Alignment have been proposed to suppress models' natural tendency to overfit degradations and yield promising results. Nevertheless, these works have assumed that models overfit to all degradation types (e.g., blur, noise, JPEG), while through careful investigations in this paper, we discover that models predominantly overfit to noise, largely attributable to its distinct degradation pattern compared to other degradation types. In this paper, we propose a targeted feature denoising framework, comprising noise detection and denoising modules. Our approach presents a general solution that can be seamlessly integrated with existing super-resolution models without requiring architectural modifications. Our framework demonstrates superior performance compared to previous regularization-based methods across five traditional benchmarks and datasets, encompassing both synthetic and real-world scenarios.

Hanyang Wang, Yuxuan Yang, Hongjun Wang et al.

The intelligent fault diagnosis of rotating mechanical equipment usually requires a large amount of labeled sample data. However, in practical industrial applications, acquiring enough data is both challenging and expensive in terms of time and cost. Moreover, different types of rotating mechanical equipment with different unique mechanical properties, require separate training of diagnostic models for each case. To address the challenges of limited fault samples and the lack of generalizability in prediction models for practical engineering applications, we propose a Multi-Attention Meta Transformer method for few-shot unsupervised rotating machinery fault diagnosis (MMT-FD). This framework extracts potential fault representations from unlabeled data and demonstrates strong generalization capabilities, making it suitable for diagnosing faults across various types of mechanical equipment. The MMT-FD framework integrates a time-frequency domain encoder and a meta-learning generalization model. The time-frequency domain encoder predicts status representations generated through random augmentations in the time-frequency domain. These enhanced data are then fed into a meta-learning network for classification and generalization training, followed by fine-tuning using a limited amount of labeled data. The model is iteratively optimized using a small number of contrastive learning iterations, resulting in high efficiency. To validate the framework, we conducted experiments on a bearing fault dataset and rotor test bench data. The results demonstrate that the MMT-FD model achieves 99\% fault diagnosis accuracy with only 1\% of labeled sample data, exhibiting robust generalization capabilities.

Qiyue Sun, Qiming Huang, Yang Yang et al.

Humans usually show exceptional generalisation and discovery ability in the open world, when being shown uncommon new concepts. Whereas most existing studies in the literature focus on common typical data from closed sets, open-world novel discovery is under-explored in videos. In this paper, we are interested in asking: What if atypical unusual videos are exposed in the learning process? To this end, we collect a new video dataset consisting of various types of unusual atypical data (e.g., sci-fi, animation, etc.). To study how such atypical data may benefit open-world learning, we feed them into the model training process for representation learning. Focusing on three key tasks in open-world learning: out-of-distribution (OOD) detection, novel category discovery (NCD), and zero-shot action recognition (ZSAR), we found that even straightforward learning approaches with atypical data consistently improve performance across various settings. Furthermore, we found that increasing the categorical diversity of the atypical samples further boosts OOD detection performance. Additionally, in the NCD task, using a smaller yet more semantically diverse set of atypical samples leads to better performance compared to using a larger but more typical dataset. In the ZSAR setting, the semantic diversity of atypical videos helps the model generalise better to unseen action classes. These observations in our extensive experimental evaluations reveal the benefits of atypical videos for visual representation learning in the open world, together with the newly proposed dataset, encouraging further studies in this direction. The project page is at: https://julysun98.github.io/atypical_dataset.

Mingze Ma, Qingtian Zhu, Yifan Zhan et al.

Implicit Neural Representations (INRs) employ neural networks to represent continuous functions by mapping coordinates to the corresponding values of the target function, with applications e.g., inverse graphics. However, INRs face a challenge known as spectral bias when dealing with scenes containing varying frequencies. To overcome spectral bias, the most common approach is the Fourier features-based methods such as positional encoding. However, Fourier features-based methods will introduce noise to output, which degrades their performances when applied to downstream tasks. In response, this paper initially hypothesizes that combining multi-layer perceptrons (MLPs) with Fourier feature embeddings mutually enhances their strengths, yet simultaneously introduces limitations inherent in Fourier feature embeddings. By presenting a simple theorem, we validate our hypothesis, which serves as a foundation for the design of our solution. Leveraging these insights, we propose the use of multi-layer perceptrons (MLPs) without additive

Hongjun Wang, Jiyuan Chen, Lun Du et al.

Recent years have witnessed the great potential of attention mechanism in graph representation learning. However, while variants of attention-based GNNs are setting new benchmarks for numerous real-world datasets, recent works have pointed out that their induced attentions are less robust and generalizable against noisy graphs due to lack of direct supervision. In this paper, we present a new framework which utilizes the tool of causality to provide a powerful supervision signal for the learning process of attention functions. Specifically, we estimate the direct causal effect of attention to the final prediction, and then maximize such effect to guide attention attending to more meaningful neighbors. Our method can serve as a plug-and-play module for any canonical attention-based GNNs in an end-to-end fashion. Extensive experiments on a wide range of benchmark datasets illustrated that, by directly supervising attention functions, the model is able to converge faster with a clearer decision boundary, and thus yields better performances.

Hongjun Wang, Guanbin Li, Xiaobai Liu et al.

Although deep convolutional neural networks (CNNs) have demonstrated remarkable performance on multiple computer vision tasks, researches on adversarial learning have shown that deep models are vulnerable to adversarial examples, which are crafted by adding visually imperceptible perturbations to the input images. Most of the existing adversarial attack methods only create a single adversarial example for the input, which just gives a glimpse of the underlying data manifold of adversarial examples. An attractive solution is to explore the solution space of the adversarial examples and generate a diverse bunch of them, which could potentially improve the robustness of real-world systems and help prevent severe security threats and vulnerabilities. In this paper, we present an effective method, called Hamiltonian Monte Carlo with Accumulated Momentum (HMCAM), aiming to generate a sequence of adversarial examples. To improve the efficiency of HMC, we propose a new regime to automatically control the length of trajectories, which allows the algorithm to move with adaptive step sizes along the search direction at different positions. Moreover, we revisit the reason for high computational cost of adversarial training under the view of MCMC and design a new generative method called Contrastive Adversarial Training (CAT), which approaches equilibrium distribution of adversarial examples with only few iterations by building from small modifications of the standard Contrastive Divergence (CD) and achieve a trade-off between efficiency and accuracy. Both quantitative and qualitative analysis on several natural image datasets and practical systems have confirmed the superiority of the proposed algorithm.

Ping Cai, Xingyuan Chen, Peng Jin et al.

The purpose of unconditional text generation is to train a model with real sentences, then generate novel sentences of the same quality and diversity as the training data. However, when different metrics are used for comparing the methods of unconditional text generation, contradictory conclusions are drawn. The difficulty is that both the diversity and quality of the sample should be considered simultaneously when the models are evaluated. To solve this problem, a novel metric of distributional discrepancy (DD) is designed to evaluate generators based on the discrepancy between the generated and real training sentences. However, it cannot compute the DD directly because the distribution of real sentences is unavailable. Thus, we propose a method for estimating the DD by training a neural-network-based text classifier. For comparison, three existing metrics, bi-lingual evaluation understudy (BLEU) versus self-BLEU, language model score versus reverse language model score, and Fréchet embedding distance, along with the proposed DD, are used to evaluate two popular generative models of long short-term memory and generative pretrained transformer 2 on both syntactic and real data. Experimental results show that DD is significantly better than the three existing metrics for ranking these generative models.

Xingyuan Chen, Ping Cai, Peng Jin et al.

The autoregressive language model (ALM) trained with maximum likelihood estimation (MLE) is widely used in unconditional text generation. Due to exposure bias, the generated texts still suffer from low quality and diversity. This presents statistically as a discrepancy between the real text and generated text. Some research shows a discriminator can detect this discrepancy. Because the discriminator can encode more information than the generator, discriminator has the potentiality to improve generator. To alleviate the exposure bias, generative adversarial networks (GAN) use the discriminator to update the generator's parameters directly, but they fail by being evaluated precisely. A critical reason for the failure is the difference between the discriminator input and the ALM input. We propose a novel mechanism by adding a filter which has the same input as the discriminator. First, discriminator detects the discrepancy signals and passes to filter directly (or by learning). Then, we use the filter to reject some generated samples with a sampling-based method. Thus, the original generative distribution is revised to reduce the discrepancy. Two ALMs, RNN-based and Transformer-based, are experimented. Evaluated precisely by three metrics, our mechanism consistently outperforms the ALMs and all kinds of GANs across two benchmark data sets.

Jielei Chu, Jing Liu, Hongjun Wang et al.

The instability is shown in the existing methods of representation learning based on Euclidean distance under a broad set of conditions. Furthermore, the scarcity and high cost of labels prompt us to explore more expressive representation learning methods which depends on the labels as few as possible. To address these issues, the small-perturbation ideology is firstly introduced on the representation learning model based on the representation probability distribution. The positive small-perturbation information (SPI) which only depend on two labels of each cluster is used to stimulate the representation probability distribution and then two variant models are proposed to fine-tune the expected representation distribution of RBM, namely, Micro-supervised Disturbance GRBM (Micro-DGRBM) and Micro-supervised Disturbance RBM (Micro-DRBM) models. The Kullback-Leibler (KL) divergence of SPI is minimized in the same cluster to promote the representation probability distributions to become more similar in Contrastive Divergence(CD) learning. In contrast, the KL divergence of SPI is maximized in the different clusters to enforce the representation probability distributions to become more dissimilar in CD learning. To explore the representation learning capability under the continuous stimulation of the SPI, we present a deep Micro-supervised Disturbance Learning (Micro-DL) framework based on the Micro-DGRBM and Micro-DRBM models and compare it with a similar deep structure which has not any external stimulation. Experimental results demonstrate that the proposed deep Micro-DL architecture shows better performance in comparison to the baseline method, the most related shallow models and deep frameworks for clustering.

Xingyuan Chen, Ping Cai, Peng Jin et al.

The text generated by neural language models is not as good as the real text. This means that their distributions are different. Generative Adversarial Nets (GAN) are used to alleviate it. However, some researchers argue that GAN variants do not work at all. When both sample quality (such as Bleu) and sample diversity (such as self-Bleu) are taken into account, the GAN variants even are worse than a well-adjusted language model. But, Bleu and self-Bleu can not precisely measure this distributional discrepancy. In fact, how to measure the distributional discrepancy between real text and generated text is still an open problem. In this paper, we theoretically propose two metric functions to measure the distributional difference between real text and generated text. Besides that, a method is put forward to estimate them. First, we evaluate language model with these two functions and find the difference is huge. Then, we try several methods to use the detected discrepancy signal to improve the generator. However the difference becomes even bigger than before. Experimenting on two existing language GANs, the distributional discrepancy between real text and generated text increases with more adversarial learning rounds. It demonstrates both of these language GANs fail.

Jielei Chu, Hongjun Wang, Jing Liu et al.

The excellent performance of representation learning of autoencoders have attracted considerable interest in various applications. However, the structure and multi-local collaborative relationships of unlabeled data are ignored in their encoding procedure that limits the capability of feature extraction. This paper presents a Multi-local Collaborative AutoEncoder (MC-AE), which consists of novel multi-local collaborative representation RBM (mcrRBM) and multi-local collaborative representation GRBM (mcrGRBM) models. Here, the Locality Sensitive Hashing (LSH) method is used to divide the input data into multi-local cross blocks which contains multi-local collaborative relationships of the unlabeled data and features since the similar multi-local instances and features of the input data are divided into the same block. In mcrRBM and mcrGRBM models, the structure and multi-local collaborative relationships of unlabeled data are integrated into their encoding procedure. Then, the local hidden features converges on the center of each local collaborative block. Under the collaborative joint influence of each local block, the proposed MC-AE has powerful capability of representation learning for unsupervised clustering. However, our MC-AE model perhaps perform training process for a long time on the large-scale and high-dimensional datasets because more local collaborative blocks are integrate into it. Five most related deep models are compared with our MC-AE. The experimental results show that the proposed MC-AE has more excellent capabilities of collaborative representation and generalization than the contrastive deep models.

Jielei Chu, Hongjun Wang, Jing Liu et al.

In this paper, we present a novel unsupervised feature learning architecture, which consists of a multi-clustering integration module and a variant of RBM termed multi-clustering integration RBM (MIRBM). In the multi-clustering integration module, we apply three unsupervised K-means, affinity propagation and spectral clustering algorithms to obtain three different clustering partitions (CPs) without any background knowledge or label. Then, an unanimous voting strategy is used to generate a local clustering partition (LCP). The novel MIRBM model is a core feature encoding part of the proposed unsupervised feature learning architecture. The novelty of it is that the LCP as an unsupervised guidance is integrated into one step contrastive divergence (CD1) learning to guide the distribution of the hidden layer features. For the instance in the same LCP cluster, the hidden and reconstructed hidden layer features of the MIRBM model in the proposed architecture tend to constrict together in the training process. Meanwhile, each LCP center tends to disperse from each other as much as possible in the hidden and reconstructed hidden layer during training. The experiments demonstrate that the proposed unsupervised feature learning architecture has more powerful feature representation and generalization capability than the state-of-the-art graph regularized RBM (GraphRBM) for clustering tasks in the Microsoft Research Asia Multimedia (MSRA-MM)2.0 dataset.

Lingbo Liu, Hongjun Wang, Guanbin Li et al.

Crowd counting from unconstrained scene images is a crucial task in many real-world applications like urban surveillance and management, but it is greatly challenged by the camera's perspective that causes huge appearance variations in people's scales and rotations. Conventional methods address such challenges by resorting to fixed multi-scale architectures that are often unable to cover the largely varied scales while ignoring the rotation variations. In this paper, we propose a unified neural network framework, named Deep Recurrent Spatial-Aware Network, which adaptively addresses the two issues in a learnable spatial transform module with a region-wise refinement process. Specifically, our framework incorporates a Recurrent Spatial-Aware Refinement (RSAR) module iteratively conducting two components: i) a Spatial Transformer Network that dynamically locates an attentional region from the crowd density map and transforms it to the suitable scale and rotation for optimal crowd estimation; ii) a Local Refinement Network that refines the density map of the attended region with residual learning. Extensive experiments on four challenging benchmarks show the effectiveness of our approach. Specifically, comparing with the existing best-performing methods, we achieve an improvement of 12% on the largest dataset WorldExpo'10 and 22.8% on the most challenging dataset UCF_CC_50.

Jielei Chu, Hongjun Wang, Hua Meng et al.

Restricted Boltzmann machines (RBMs) and their variants are usually trained by contrastive divergence (CD) learning, but the training procedure is an unsupervised learning approach, without any guidances of the background knowledge. To enhance the expression ability of traditional RBMs, in this paper, we propose pairwise constraints restricted Boltzmann machine with Gaussian visible units (pcGRBM) model, in which the learning procedure is guided by pairwise constraints and the process of encoding is conducted under these guidances. The pairwise constraints are encoded in hidden layer features of pcGRBM. Then, some pairwise hidden features of pcGRBM flock together and another part of them are separated by the guidances. In order to deal with real-valued data, the binary visible units are replaced by linear units with Gausian noise in the pcGRBM model. In the learning process of pcGRBM, the pairwise constraints are iterated transitions between visible and hidden units during CD learning procedure. Then, the proposed model is inferred by approximative gradient descent method and the corresponding learning algorithm is designed in this paper. In order to compare the availability of pcGRBM and traditional RBMs with Gaussian visible units, the features of the pcGRBM and RBMs hidden layer are used as input 'data' for K-means, spectral clustering (SP) and affinity propagation (AP) algorithms, respectively. A thorough experimental evaluation is performed with sixteen image datasets of Microsoft Research Asia Multimedia (MSRA-MM). The experimental results show that the clustering performance of K-means, SP and AP algorithms based on pcGRBM model are significantly better than traditional RBMs. In addition, the pcGRBM model for clustering task shows better performance than some semi-supervised clustering algorithms.