Hao Zheng

Wei Guo, Yuanhao Li, Hao Zheng et al.

As cellular core networks evolve toward distributed and cloud-native architectures, control-plane interactions become more intricate and bring new challenges. Among these challenges, we find that introducing specific timing between two control-plane interactions can cause network function crash, which we define as timing-induced interaction failures. Prior research primarily addresses identifying malformed inputs and specification violations, while timing-induced interaction failures remain largely unexplored. In this paper, we conduct a systematic study of timing-induced interaction failures in LTE and 5G core networks. First, we establish a taxonomy of control-plane interaction patterns and analyze the failure modes of each pattern. Then, we design and implement Kairos, a lightweight testing framework to expose timing-induced interaction failures without analyzing cellular standard documents. Evaluating Kairos on two open source and two commercial LTE and 5G core networks, we uncover 20 new vulnerabilities and reproduce 34 existing issues. Our results show that timing-induced interaction failures are prevalent in LTE and 5G core networks and should be explicitly considered in future specifications.

Hao Zheng, Hu Wang, Tiantian Zheng et al.

Dual-hand action segmentation, densely predicting actions for both hands from untrimmed videos, is essential for understanding complex bimanual activities. However, it poses several unique challenges: complex inter-hand dependencies, visual asymmetry between hands, representation conflicts where the dominant hand monopolizes gradients, and semantic ambiguity in fine-grained actions. We propose Polyphony, a three-stage method to address these challenges through: (1) an Alternating Dual-Hand Vision Transformer that alternates training between left- and right-hand mini-batches to ensure balanced gradient contributions from both hands while sharing a spatio-temporal encoder; (2) Semantic Feature Conditioning that aligns visual features with structured, compositional action descriptions to enhance discrimination of semantically similar actions; and (3) Diffusion-Based Segmentation with cross-hand feature fusion for inter-hand coordination and adaptive loss weighting for balancing performance. Polyphony achieves state-of-the-art on both dual-hand datasets (HA-ViD, ATTACH) with improvements up to 16.8 points, and on the single-stream Breakfast dataset (82.5%), outperforming the prior best method that uses a 12x larger backbone. Notably, our unified model with a single shared backbone surpasses baselines requiring separate per-hand models. Code is at https://github.com/x-labs-xyz/Polyphony-Dual-hand-Action-Segmentation.

Hao Zheng, Runqi Wang, Jianzhuang Liu et al.

The conventional few-shot classification aims at learning a model on a large labeled base dataset and rapidly adapting to a target dataset that is from the same distribution as the base dataset. However, in practice, the base and the target datasets of few-shot classification are usually from different domains, which is the problem of cross-domain few-shot classification. We tackle this problem by making a small proportion of unlabeled images in the target domain accessible in the training stage. In this setup, even though the base data are sufficient and labeled, the large domain shift still makes transferring the knowledge from the base dataset difficult. We meticulously design a cross-level knowledge distillation method, which can strengthen the ability of the model to extract more discriminative features in the target dataset by guiding the network's shallow layers to learn higher-level information. Furthermore, in order to alleviate the overfitting in the evaluation stage, we propose a feature denoising operation which can reduce the feature redundancy and mitigate overfitting. Our approach can surpass the previous state-of-the-art method, Dynamic-Distillation, by 5.44% on 1-shot and 1.37% on 5-shot classification tasks on average in the BSCD-FSL benchmark. The implementation code will be available at https://github.com/jarucezh/cldfd.

Yizhe Zhang, Suraj Mishra, Peixian Liang et al.

We aim to quantitatively measure the practical usability of medical image segmentation models: to what extent, how often, and on which samples a model's predictions can be used/trusted. We first propose a measure, Correctness-Confidence Rank Correlation (CCRC), to capture how predictions' confidence estimates correlate with their correctness scores in rank. A model with a high value of CCRC means its prediction confidences reliably suggest which samples' predictions are more likely to be correct. Since CCRC does not capture the actual prediction correctness, it alone is insufficient to indicate whether a prediction model is both accurate and reliable to use in practice. Therefore, we further propose another method, Usable Region Estimate (URE), which simultaneously quantifies predictions' correctness and reliability of confidence assessments in one estimate. URE provides concrete information on to what extent a model's predictions are usable. In addition, the sizes of usable regions (UR) can be utilized to compare models: A model with a larger UR can be taken as a more usable and hence better model. Experiments on six datasets validate that the proposed evaluation methods perform well, providing a concrete and concise measure for the practical usability of medical image segmentation models. Code is made available at https://github.com/yizhezhang2000/ure.

Minghui Zhang, Yangqian Wu, Hanxiao Zhang et al. · harvard

Open international challenges are becoming the de facto standard for assessing computer vision and image analysis algorithms. In recent years, new methods have extended the reach of pulmonary airway segmentation that is closer to the limit of image resolution. Since EXACT'09 pulmonary airway segmentation, limited effort has been directed to quantitative comparison of newly emerged algorithms driven by the maturity of deep learning based approaches and clinical drive for resolving finer details of distal airways for early intervention of pulmonary diseases. Thus far, public annotated datasets are extremely limited, hindering the development of data-driven methods and detailed performance evaluation of new algorithms. To provide a benchmark for the medical imaging community, we organized the Multi-site, Multi-domain Airway Tree Modeling (ATM'22), which was held as an official challenge event during the MICCAI 2022 conference. ATM'22 provides large-scale CT scans with detailed pulmonary airway annotation, including 500 CT scans (300 for training, 50 for validation, and 150 for testing). The dataset was collected from different sites and it further included a portion of noisy COVID-19 CTs with ground-glass opacity and consolidation. Twenty-three teams participated in the entire phase of the challenge and the algorithms for the top ten teams are reviewed in this paper. Quantitative and qualitative results revealed that deep learning models embedded with the topological continuity enhancement achieved superior performance in general. ATM'22 challenge holds as an open-call design, the training data and the gold standard evaluation are available upon successful registration via its homepage.

Chuyan Zhang, Yuncheng Yang, Hao Zheng et al.

Driven by the latest trend towards self-supervised learning (SSL), the paradigm of "pretraining-then-finetuning" has been extensively explored to enhance the performance of clinical applications with limited annotations. Previous literature on model finetuning has mainly focused on regularization terms and specific policy models, while the misalignment of channels between source and target models has not received sufficient attention. In this work, we revisited the dynamics of batch normalization (BN) layers and observed that the trainable affine parameters of BN serve as sensitive indicators of domain information. Therefore, Affine Collaborative Normalization (AC-Norm) is proposed for finetuning, which dynamically recalibrates the channels in the target model according to the cross-domain channel-wise correlations without adding extra parameters. Based on a single-step backpropagation, AC-Norm can also be utilized to measure the transferability of pretrained models. We evaluated AC-Norm against the vanilla finetuning and state-of-the-art fine-tuning methods on transferring diverse pretrained models to the diabetic retinopathy grade classification, retinal vessel segmentation, CT lung nodule segmentation/classification, CT liver-tumor segmentation and MRI cardiac segmentation tasks. Extensive experiments demonstrate that AC-Norm unanimously outperforms the vanilla finetuning by up to 4% improvement, even under significant domain shifts where the state-of-the-art methods bring no gains. We also prove the capability of AC-Norm in fast transferability estimation. Our code is available at https://github.com/EndoluminalSurgicalVision-IMR/ACNorm.

Hao Zheng, Guozhao Mo, Xinru Yan et al.

Presentation generation requires deep content research, coherent visual design, and iterative refinement based on observation. However, existing presentation agents often rely on predefined workflows and fixed templates. To address this, we present DeepPresenter, an agentic framework that adapts to diverse user intents, enables effective feedback-driven refinement, and generalizes beyond a scripted pipeline. Specifically, DeepPresenter autonomously plans, renders, and revises intermediate slide artifacts to support long-horizon refinement with environmental observations. Furthermore, rather than relying on self-reflection over internal signals (e.g., reasoning traces), our environment-grounded reflection conditions the generation process on perceptual artifact states (e.g., rendered slides), enabling the system to identify and correct presentation-specific issues during execution. Results on the evaluation set covering diverse presentation-generation scenarios show that DeepPresenter achieves state-of-the-art performance, and the fine-tuned 9B model remains highly competitive at substantially lower cost. Our project is available at: https://github.com/icip-cas/PPTAgent

Hanbo Huang, Yiran Zhang, Hao Zheng et al.

Large language model (LLM) watermarking has shown promise in detecting AI-generated content and mitigating misuse, with prior work claiming robustness against paraphrasing and text editing. In this paper, we argue that existing evaluations are not sufficiently adversarial, obscuring critical vulnerabilities and overstating the security. To address this, we introduce the adaptive robustness radius, a formal metric that quantifies the worst-case resilience of watermarks against adaptive adversaries. By lifting the paraphrase space into a KL-divergence ball, we approximate this radius and theoretically demonstrate that optimizing the attack context and model parameters can significantly reduce the approximate radius, making watermarks highly vulnerable to paraphrase attacks. Leveraging this insight, we propose RLCracker, a reinforcement learning (RL)-based adaptive attack that erases watermark signals with limited watermarked examples and limited access to the detector. Despite weak supervision, it empowers a 3B model to achieve 98.5% removal success with minimal semantic shift on 1,500-token Unigram-marked texts after training on only 100 short samples. This performance dramatically exceeds 6.75% by GPT-4o and generalizes across five model sizes over ten watermarking schemes. Our code is available at https://github.com/OTT0-OTO/RLCracker.

Hao Zheng, Mirella Lapata

Compositional generalization is a basic mechanism in human language learning, which current neural networks struggle with. A recently proposed Disentangled sequence-to-sequence model (Dangle) shows promising generalization capability by learning specialized encodings for each decoding step. We introduce two key modifications to this model which encourage more disentangled representations and improve its compute and memory efficiency, allowing us to tackle compositional generalization in a more realistic setting. Specifically, instead of adaptively re-encoding source keys and values at each time step, we disentangle their representations and only re-encode keys periodically, at some interval. Our new architecture leads to better generalization performance across existing tasks and datasets, and a new machine translation benchmark which we create by detecting naturally occurring compositional patterns in relation to a training set. We show this methodology better emulates real-world requirements than artificial challenges.

Conghao Xiong, Yi Lin, Hao Chen et al.

Transferring knowledge from a source domain to a target domain can be crucial for whole slide image classification, since the number of samples in a dataset is often limited due to high annotation costs. However, domain shift and task discrepancy between datasets can hinder effective knowledge transfer. In this paper, we propose a Target-Aware Knowledge Transfer framework, employing a teacher-student paradigm. Our framework enables the teacher model to learn common knowledge from the source and target domains by actively incorporating unlabelled target images into the training of the teacher model. The teacher bag features are subsequently adapted to supervise the training of the student model on the target domain. Despite incorporating the target features during training, the teacher model tends to overlook them under the inherent domain shift and task discrepancy. To alleviate this, we introduce a target-aware feature alignment module to establish a transferable latent relationship between the source and target features by solving the optimal transport problem. Experimental results show that models employing knowledge transfer outperform those trained from scratch, and our method achieves state-of-the-art performance among other knowledge transfer methods on various datasets, including TCGA-RCC, TCGA-NSCLC, and Camelyon16.

Hao Zheng, Regina Lee, Yuqian Lu

Understanding comprehensive assembly knowledge from videos is critical for futuristic ultra-intelligent industry. To enable technological breakthrough, we present HA-ViD - the first human assembly video dataset that features representative industrial assembly scenarios, natural procedural knowledge acquisition process, and consistent human-robot shared annotations. Specifically, HA-ViD captures diverse collaboration patterns of real-world assembly, natural human behaviors and learning progression during assembly, and granulate action annotations to subject, action verb, manipulated object, target object, and tool. We provide 3222 multi-view, multi-modality videos (each video contains one assembly task), 1.5M frames, 96K temporal labels and 2M spatial labels. We benchmark four foundational video understanding tasks: action recognition, action segmentation, object detection and multi-object tracking. Importantly, we analyze their performance for comprehending knowledge in assembly progress, process efficiency, task collaboration, skill parameters and human intention. Details of HA-ViD is available at: https://iai-hrc.github.io/ha-vid.

Xiaoyang Chen, Hao Zheng, Yuemeng Li et al.

A versatile medical image segmentation model applicable to images acquired with diverse equipment and protocols can facilitate model deployment and maintenance. However, building such a model typically demands a large, diverse, and fully annotated dataset, which is challenging to obtain due to the labor-intensive nature of data curation. To address this challenge, we propose a cost-effective alternative that harnesses multi-source data with only partial or sparse segmentation labels for training, substantially reducing the cost of developing a versatile model. We devise strategies for model self-disambiguation, prior knowledge incorporation, and imbalance mitigation to tackle challenges associated with inconsistently labeled multi-source data, including label ambiguity and modality, dataset, and class imbalances. Experimental results on a multi-modal dataset compiled from eight different sources for abdominal structure segmentation have demonstrated the effectiveness and superior performance of our method compared to state-of-the-art alternative approaches. We anticipate that its cost-saving features, which optimize the utilization of existing annotated data and reduce annotation efforts for new data, will have a significant impact in the field.

Jingjun Yi, Qi Bi, Hao Zheng et al.

The rapid development of Vision Foundation Model (VFM) brings inherent out-domain generalization for a variety of down-stream tasks. Among them, domain generalized semantic segmentation (DGSS) holds unique challenges as the cross-domain images share common pixel-wise content information but vary greatly in terms of the style. In this paper, we present a novel Spectral-dEcomposed Token (SET) learning framework to advance the frontier. Delving into further than existing fine-tuning token & frozen backbone paradigm, the proposed SET especially focuses on the way learning style-invariant features from these learnable tokens. Particularly, the frozen VFM features are first decomposed into the phase and amplitude components in the frequency space, which mainly contain the information of content and style, respectively, and then separately processed by learnable tokens for task-specific information extraction. After the decomposition, style variation primarily impacts the token-based feature enhancement within the amplitude branch. To address this issue, we further develop an attention optimization method to bridge the gap between style-affected representation and static tokens during inference. Extensive cross-domain experiments show its state-of-the-art performance.

Hao Zheng, Hui Lin, Rong Zhao et al.

The binding problem is one of the fundamental challenges that prevent the artificial neural network (ANNs) from a compositional understanding of the world like human perception, because disentangled and distributed representations of generative factors can interfere and lead to ambiguity when complex data with multiple objects are presented. In this paper, we propose a brain-inspired hybrid neural network (HNN) that introduces temporal binding theory originated from neuroscience into ANNs by integrating spike timing dynamics (via spiking neural networks, SNNs) with reconstructive attention (by ANNs). Spike timing provides an additional dimension for grouping, while reconstructive feedback coordinates the spikes into temporal coherent states. Through iterative interaction of ANN and SNN, the model continuously binds multiple objects at alternative synchronous firing times in the SNN coding space. The effectiveness of the model is evaluated on synthetic datasets of binary images. By visualization and analysis, we demonstrate that the binding is explainable, soft, flexible, and hierarchical. Notably, the model is trained on single object datasets without explicit supervision on grouping, but successfully binds multiple objects on test datasets, showing its compositional generalization capability. Further results show its binding ability in dynamic situations.

Haoxuan Jia, Yang Liu, Bin Chong et al.

Finance LLM agents must simultaneously block prompt-induced unauthorized actions and approve legitimate multi-step business workflows. However, boundary filters often miss irreversible mid-trajectory tool calls, while post-hoc LLM judges perform auditing only after termination -- too late for intervention and at a computational cost that scales linearly with trace length. We present FinHarness, an inline safety harness that wraps a finance agent end-to-end with three components: a Query Monitor that fuses single-turn intent with cross-turn drift, a Tool Monitor that evaluates each prospective tool call, and a Cascade module that integrates per-step risk and adaptively routes verification between a lightweight and an advanced-tier LLM judge. Fired risk factors are re-injected into the agent input as ex-ante evidence, enabling the agent to refuse, re-plan, or approve on its own. On FinVault, routed FinHarness cuts ASR from 38.3% to 15.0% while largely preserving benign approval ($41.1\% \to 39.3\%$), and uses $4.7\times$ fewer advanced-judge calls than an always-advanced ablation.

Hao Zheng, Hongming Li, Yong Fan

To achieve fast, robust, and accurate reconstruction of the human cortical surfaces from 3D magnetic resonance images (MRIs), we develop a novel deep learning-based framework, referred to as SurfNN, to reconstruct simultaneously both inner (between white matter and gray matter) and outer (pial) surfaces from MRIs. Different from existing deep learning-based cortical surface reconstruction methods that either reconstruct the cortical surfaces separately or neglect the interdependence between the inner and outer surfaces, SurfNN reconstructs both the inner and outer cortical surfaces jointly by training a single network to predict a midthickness surface that lies at the center of the inner and outer cortical surfaces. The input of SurfNN consists of a 3D MRI and an initialization of the midthickness surface that is represented both implicitly as a 3D distance map and explicitly as a triangular mesh with spherical topology, and its output includes both the inner and outer cortical surfaces, as well as the midthickness surface. The method has been evaluated on a large-scale MRI dataset and demonstrated competitive cortical surface reconstruction performance.

Bojian Hou, Hongming Li, Zhicheng Jiao et al.

Conventional survival analysis methods are typically ineffective to characterize heterogeneity in the population while such information can be used to assist predictive modeling. In this study, we propose a hybrid survival analysis method, referred to as deep clustering survival machines, that combines the discriminative and generative mechanisms. Similar to the mixture models, we assume that the timing information of survival data is generatively described by a mixture of certain numbers of parametric distributions, i.e., expert distributions. We learn weights of the expert distributions for individual instances according to their features discriminatively such that each instance's survival information can be characterized by a weighted combination of the learned constant expert distributions. This method also facilitates interpretable subgrouping/clustering of all instances according to their associated expert distributions. Extensive experiments on both real and synthetic datasets have demonstrated that the method is capable of obtaining promising clustering results and competitive time-to-event predicting performance.

Md Rubel Ahmed, Bardia Nadimi, Hao Zheng

High-quality system-level message flow specifications are necessary for comprehensive validation of system-on-chip (SoC) designs. However, manual development and maintenance of such specifications are daunting tasks. We propose a disruptive method that utilizes deep sequence modeling with the attention mechanism to infer accurate flow specifications from SoC communication traces. The proposed method can overcome the inherent complexity of SoC traces induced by the concurrent executions of SoC designs that existing mining tools often find extremely challenging. We conduct experiments on five highly concurrent traces and find that the proposed approach outperforms several existing state-of-the-art trace mining tools.

Md Rubel Ahmed, Hao Zheng

High-quality system-level message flow specifications can lead to comprehensive validation of system-on-chip (SoC) designs. We propose a disruptive method that utilizes an attention mechanism to produce accurate flow specifications from SoC IP communication traces. The proposed method can overcome the inherent complexity of SoC traces induced by the concurrency and parallelism of multicore designs that existing flow specification mining tools often find extremely challenging. Experiments on highly interleaved traces show promising flow reconstruction compared to several tools dedicated to the flow specification mining problem.

Jaime Rafael Imperial, Hao Zheng

Existing Large Language Model (LLM) approaches to SystemVerilog Assertion (SVA) generation primarily focus on syntactic validity and formal verification outcomes, while semantic alignment between generated assertions and natural language specifications remains difficult to quantify. As a result, hallucinated or misaligned SVAs can reduce confidence and increase debugging efforts in the absence of golden RTL. This paper presents SpecAlign, a framework for semantic evaluation and refinement of LLM-generated SVAs. SpecAlign introduces two iterative alignment loops that assess both natural language properties and SVAs against the design specification using entailment-based classification. We improve alignment decisions by generating multiple reasoning paths using chain-of-thought prompting and aggregating them via a self-consistency voting mechanism. Misaligned assertions are analyzed to generate actionable feedback for refinement. We further define a quantitative alignment score to measure semantic consistency across iterations. Experimental results demonstrate that SpecAlign effectively detects semantic inconsistencies and improves assertion alignment without relying on golden RTL, providing a scalable complement to traditional formal verification evaluation metrics.

Xueming Fu, Hao Zheng, Luyan Liu et al.

Predicting lower limb motion intent is vital for controlling exoskeleton robots and prosthetic limbs. Surface electromyography (sEMG) attracts increasing attention in recent years as it enables ahead-of-time prediction of motion intentions before actual movement. However, the estimation performance of human joint trajectory remains a challenging problem due to the inter- and intra-subject variations. The former is related to physiological differences (such as height and weight) and preferred walking patterns of individuals, while the latter is mainly caused by irregular and gait-irrelevant muscle activity. This paper proposes a model integrating two gait cycle-inspired learning strategies to mitigate the challenge for predicting human knee joint trajectory. The first strategy is to decouple knee joint angles into motion patterns and amplitudes former exhibit low variability while latter show high variability among individuals. By learning through separate network entities, the model manages to capture both the common and personalized gait features. In the second, muscle principal activation masks are extracted from gait cycles in a prolonged walk. These masks are used to filter out components unrelated to walking from raw sEMG and provide auxiliary guidance to capture more gait-related features. Experimental results indicate that our model could predict knee angles with the average root mean square error (RMSE) of 3.03(0.49) degrees and 50ms ahead of time. To our knowledge this is the best performance in relevant literatures that has been reported, with reduced RMSE by at least 9.5%.

Junjie Yu, Pengrui Lu, Weiye Si et al.

Benchmarks within the OpenClaw ecosystem have thus far evaluated exclusively assistant-level tasks, leaving the academic-level capabilities of OpenClaw largely unexamined. We introduce AcademiClaw, a bilingual benchmark of 80 complex, long-horizon tasks sourced directly from university students' real academic workflows -- homework, research projects, competitions, and personal projects -- that they found current AI agents unable to solve effectively. Curated from 230 student-submitted candidates through rigorous expert review, the final task set spans 25+ professional domains, ranging from olympiad-level mathematics and linguistics problems to GPU-intensive reinforcement learning and full-stack system debugging, with 16 tasks requiring CUDA GPU execution. Each task executes in an isolated Docker sandbox and is scored on task completion by multi-dimensional rubrics combining six complementary techniques, with an independent five-category safety audit providing additional behavioral analysis. Experiments on six frontier models show that even the best achieves only a 55\% pass rate. Further analysis uncovers sharp capability boundaries across task domains, divergent behavioral strategies among models, and a disconnect between token consumption and output quality, providing fine-grained diagnostic signals beyond what aggregate metrics reveal. We hope that AcademiClaw and its open-sourced data and code can serve as a useful resource for the OpenClaw community, driving progress toward agents that are more capable and versatile across the full breadth of real-world academic demands. All data and code are available at https://github.com/GAIR-NLP/AcademiClaw.

Yejia Zhang, Nishchal Sapkota, Pengfei Gu et al.

Understanding of spatial attributes is central to effective 3D radiology image analysis where crop-based learning is the de facto standard. Given an image patch, its core spatial properties (e.g., position & orientation) provide helpful priors on expected object sizes, appearances, and structures through inherent anatomical consistencies. Spatial correspondences, in particular, can effectively gauge semantic similarities between inter-image regions, while their approximate extraction requires no annotations or overbearing computational costs. However, recent 3D contrastive learning approaches either neglect correspondences or fail to maximally capitalize on them. To this end, we propose an extensible 3D contrastive framework (Spade, for Spatial Debiasing) that leverages extracted correspondences to select more effective positive & negative samples for representation learning. Our method learns both globally invariant and locally equivariant representations with downstream segmentation in mind. We also propose separate selection strategies for global & local scopes that tailor to their respective representational requirements. Compared to recent state-of-the-art approaches, Spade shows notable improvements on three downstream segmentation tasks (CT Abdominal Organ, CT Heart, MR Heart).

Jiaqi Xue, Yancheng Zhang, Yanshan Wang et al.

Secure training, while protecting the confidentiality of both data and model weights, typically incurs significant training overhead. Traditional Fully Homomorphic Encryption (FHE)-based non-inter-active training models are heavily burdened by computationally demanding bootstrapping. To develop an efficient secure training system, we established a foundational framework, CryptoTrain-B, utilizing a hybrid cryptographic protocol that merges FHE with Oblivious Transfer (OT) for handling linear and non-linear operations, respectively. This integration eliminates the need for costly bootstrapping. Although CryptoTrain-B sets a new baseline in performance, reducing its training overhead remains essential. We found that ciphertext-ciphertext multiplication (CCMul) is a critical bottleneck in operations involving encrypted inputs and models. Our solution, the CCMul-Precompute technique, involves precomputing CCMul offline and resorting to the less resource-intensive ciphertext-plaintext multiplication (CPMul) during private training. Furthermore, conventional polynomial convolution in FHE systems tends to encode irrelevant and redundant values into polynomial slots, necessitating additional polynomials and ciphertexts for input representation and leading to extra multiplications. Addressing this, we introduce correlated polynomial convolution, which encodes only related input values into polynomials, thus drastically reducing the number of computations and overheads. By integrating CCMul-Precompute and correlated polynomial convolution into CryptoTrain-B, we facilitate a rapid and efficient secure training framework, CryptoTrain. Extensive experiments demonstrate that CryptoTrain achieves a ~5.3X training time reduction compared to prior methods.

Xuwen Zhou, Fangxin Liu, Chao Wang et al.

Speculative decoding accelerates autoregressive generation by letting draft tokens bypass full verification, but conventional frameworks suffer from frequent false rejections, particularly when draft models produce semantically correct but lexically divergent outputs. In this paper, we present Calibrated Speculative Decoding (CSD), a training-free framework that recovers valid tokens discarded by standard verification. Guided by the principle of "Frequency-Guided Candidate Selection and Probability-Guarded Acceptance," CSD incorporates two lightweight modules: Online Correction Memory, which aggregates historical rejections to propose recurring divergence patterns as rescue candidates, and Semantic Consistency Gating, which verifies candidate admissibility using probability ratios instead of exact token matching. Our evaluation across diverse large language models demonstrates that CSD outperforms existing methods, achieving a peak throughput speedup of 2.33x. CSD preserves model accuracy across all tasks while further boosting performance on complex reasoning datasets. These results establish CSD as a highly effective, lightweight solution for practical LLM deployments.

Bo Xu, Hao Zheng, Zhigang Hu et al.

In current synthetic aperture radar (SAR) object classification, one of the major challenges is the severe overfitting issue due to the limited dataset (few-shot) and noisy data. Considering the advantages of knowledge distillation as a learned label smoothing regularization, this paper proposes a novel Double Reverse Regularization Network based on Self-Knowledge Distillation (DRRNet-SKD). Specifically, through exploring the effect of distillation weight on the process of distillation, we are inspired to adopt the double reverse thought to implement an effective regularization network by combining offline and online distillation in a complementary way. Then, the Adaptive Weight Assignment (AWA) module is designed to adaptively assign two reverse-changing weights based on the network performance, allowing the student network to better benefit from both teachers. The experimental results on OpenSARShip and FUSAR-Ship demonstrate that DRRNet-SKD exhibits remarkable performance improvement on classical CNNs, outperforming state-of-the-art self-knowledge distillation methods.

Yejia Zhang, Pengfei Gu, Nishchal Sapkota et al.

High annotation costs and limited labels for dense 3D medical imaging tasks have recently motivated an assortment of 3D self-supervised pretraining methods that improve transfer learning performance. However, these methods commonly lack spatial awareness despite its centrality in enabling effective 3D image analysis. More specifically, position, scale, and orientation are not only informative but also automatically available when generating image crops for training. Yet, to date, no work has proposed a pretext task that distills all key spatial features. To fulfill this need, we develop a new self-supervised method, VectorPOSE, which promotes better spatial understanding with two novel pretext tasks: Vector Prediction (VP) and Boundary-Focused Reconstruction (BFR). VP focuses on global spatial concepts (i.e., properties of 3D patches) while BFR addresses weaknesses of recent reconstruction methods to learn more effective local representations. We evaluate VectorPOSE on three 3D medical image segmentation tasks, showing that it often outperforms state-of-the-art methods, especially in limited annotation settings.

Wenhao Zhou, Hao Zheng, Rong Zhao

Large Vision-Language Models (LVLMs) typically align visual features from an encoder with a pre-trained Large Language Model (LLM). However, this makes the visual perception module a bottleneck, which constrains the overall capabilities of LVLMs. Conventional evaluation benchmarks, while rich in visual semantics, often contain unavoidable local shortcuts that can lead to an overestimation of models' perceptual abilities. Here, we introduce TopoPerception, a benchmark that leverages topological properties to rigorously evaluate the global visual perception capabilities of LVLMs across various granularities. Since topology depends on the global structure of an image and is invariant to local features, TopoPerception enables a shortcut-free assessment of global perception, fundamentally distinguishing it from semantically rich tasks. We evaluate state-of-the-art models on TopoPerception and find that even at the coarsest perceptual granularity, all models perform no better than random chance, indicating a profound inability to perceive global visual features. Notably, a consistent trend emerge within model families: more powerful models with stronger reasoning capabilities exhibit lower accuracy. This suggests that merely scaling up models is insufficient to address this deficit and may even exacerbate it. Progress may require new training paradigms or architectures. TopoPerception not only exposes a critical bottleneck in current LVLMs but also offers a lens and direction for improving their global visual perception. The data and code are publicly available at: https://github.com/Wenhao-Zhou/TopoPerception.

Xianting Feng, Hao zheng, Zhigang Hu et al.

Most existing synthetic aperture radar (SAR) ship classification technologies heavily rely on correctly labeled data, ignoring the discriminative features of unlabeled SAR ship images. Even though researchers try to enrich CNN-based features by introducing traditional handcrafted features, existing methods easily cause information redundancy and fail to capture the interaction between them. To address these issues, we propose a novel dual-stream contrastive predictive network (DCPNet), which consists of two asymmetric task designs and the false negative sample elimination module. The first task is to construct positive sample pairs, guiding the core encoder to learn more general representations. The second task is to encourage adaptive capture of the correspondence between deep features and handcrated features, achieving knowledge transfer within the model, and effectively improving the redundancy caused by the feature fusion. To increase the separability between clusters, we also design a cluster-level tasks. The experimental results on OpenSARShip and FUSAR-Ship datasets demonstrate the improvement in classification accuracy of supervised models and confirm the capability of learning effective representations of DCPNet.

Bardia Nadimi, Ghali Omar Boutaib, Hao Zheng

Recently, there has been a growing interest in leveraging Large Language Models for Verilog code generation. However, the current quality of the generated Verilog code remains suboptimal. This is largely due to the absence of well-defined, well-organized datasets with high-quality samples, as well as a lack of innovative fine-tuning methods and models specifically trained on Verilog. In this paper, we introduce a novel open-source dataset and a corresponding fine-tuning technique, which utilizes a multi-layered structure that we refer to as PyraNet. Our experiments demonstrate that employing the proposed dataset and fine-tuning approach leads to a more accurate fine-tuned model, producing syntactically and functionally correct Verilog code. The evaluation results show improvements by up-to $32.6\%$ in comparison to the CodeLlama-7B baseline model and up-to $16.7\%$ in comparison to the state-of-the-art models using VerilogEval evaluation platform.

Guijia Zhang, Hao Zheng, Harry Yang

Multimodal agents use screenshots, documents, and webpages to choose tool calls. When a false visual claim triggers a click, email, extraction, or transfer, hallucination becomes an authorization failure rather than an answer-quality error. We formalize this failure mode as hallucination-to-action conversion: an unsupported perceptual claim supplies the precondition that makes a privileged action appear permitted. We propose evidence-carrying multimodal agents (ECA), which treat free-form model text as inadmissible evidence. ECA decomposes each tool call into action-critical predicates, obtains typed certificates from constrained DOM/OCR/AX verifiers, and lets a deterministic gate grant only the privileges those certificates support. The architecture does not hide perception error; it converts opaque model belief into named verifier, schema, and implementation residuals. Verifier red-teaming over 1,900 attacks exposes this residual directly: four targeted hardening steps reduce gate bypass from 15% to 1.3%. With content-derived certificates, ECA obtains 0% unsafe-action rate on a 200-task end-to-end pipeline (Wilson 95% upper bound 2.67%) and a 120-task browser proof-of-concept (upper bound 4.3%). A direct HACR audit on 500 stratified task keys shows that unsupported action-critical claims reach unsafe execution for naive agents (100.0%) and prompt-only defense (49.6%), but not for ECA. Oracle-certificate replay on 7,488 GPT-5.4 benchmark traces serves as a gate-correctness sanity check, and neural judge baselines remain bypassable under the same threat model. The resulting principle is simple: model language may propose actions, but external evidence must authorize them.

Xinquan Yang, Jinheng Xie, Yawen Huang et al.

Long-tailed pulmonary anomalies in chest radiography present formidable diagnostic challenges. Despite the recent strides in diffusion-based methods for enhancing the representation of tailed lesions, the paucity of rare lesion exemplars curtails the generative capabilities of these approaches, thereby leaving the diagnostic precision less than optimal. In this paper, we propose a novel data synthesis pipeline designed to augment tail lesions utilizing a copious supply of conventional normal X-rays. Specifically, a sufficient quantity of normal samples is amassed to train a diffusion model capable of generating normal X-ray images. This pre-trained diffusion model is subsequently utilized to inpaint the head lesions present in the diseased X-rays, thereby preserving the tail classes as augmented training data. Additionally, we propose the integration of a Large Language Model Knowledge Guidance (LKG) module alongside a Progressive Incremental Learning (PIL) strategy to stabilize the inpainting fine-tuning process. Comprehensive evaluations conducted on the public lung datasets MIMIC and CheXpert demonstrate that the proposed method sets a new benchmark in performance.

Gheorghe Comanici, Eric Bieber, Mike Schaekermann et al. · amazon-science, baidu

In this report, we introduce the Gemini 2.X model family: Gemini 2.5 Pro and Gemini 2.5 Flash, as well as our earlier Gemini 2.0 Flash and Flash-Lite models. Gemini 2.5 Pro is our most capable model yet, achieving SoTA performance on frontier coding and reasoning benchmarks. In addition to its incredible coding and reasoning skills, Gemini 2.5 Pro is a thinking model that excels at multimodal understanding and it is now able to process up to 3 hours of video content. Its unique combination of long context, multimodal and reasoning capabilities can be combined to unlock new agentic workflows. Gemini 2.5 Flash provides excellent reasoning abilities at a fraction of the compute and latency requirements and Gemini 2.0 Flash and Flash-Lite provide high performance at low latency and cost. Taken together, the Gemini 2.X model generation spans the full Pareto frontier of model capability vs cost, allowing users to explore the boundaries of what is possible with complex agentic problem solving.

Zeyao Ma, Jing Zhang, Xiaokang Zhang et al.

Large language models (LLMs) have demonstrated strong coding capabilities but still struggle to solve competitive programming problems correctly in a single attempt. Execution-based re-ranking offers a promising test-time scaling strategy, yet existing methods are constrained by either difficult test case generation or inefficient random input sampling. To address this limitation, we propose Agentic Verifier, an execution-based agent that actively reasons about program behaviors and searches for highly discriminative test inputs that expose behavioral discrepancies among candidate solutions. Through multi-turn interaction with code execution environments, the verifier iteratively refines the candidate input generator and produces targeted counterexamples rather than blindly sampling inputs. We train the verifier to acquire this discriminative input generation capability via a scalable pipeline combining large-scale data synthesis, rejection fine-tuning, and agentic reinforcement learning. Extensive experiments across five competitive programming benchmarks demonstrate consistent improvements over strong execution-based baselines, achieving up to +10-15% absolute gains in Best@K accuracy. Further analysis reveals clear test-time scaling behavior and highlights the verifier's broader potential beyond reranking.

Jiahe Song, Chuang Wang, Yinfan Wang et al.

Reaction diagram parsing (RxnDP) is critical for extracting chemical synthesis information from literature. Although recent Vision-Language Models (VLMs) have emerged as a promising paradigm to automate this complex visual reasoning task, their application is fundamentally bottlenecked by the inability to align visual chemical entities with pre-trained knowledge, alongside the inherent discrepancy between token-level training and reaction-level evaluation. To address these dual challenges, this work enhances VLM-based RxnDP from two complementary perspectives: prompting representation and learning paradigms. First, we propose Identifier as Visual Prompting (IdtVP), which leverages naturally occurring molecule identifiers (e.g., bold numerals like 1a) to activate the chemical knowledge acquired during VLM pre-training. IdtVP enables powerful zero-shot and out-of-distribution capabilities, outperforming existing prompting strategies. Second, to further optimize performance within fine-tuning paradigms, we introduce Re3-DAPO, a reinforcement learning algorithm that leverages verifiable rewards to directly optimize reaction-level metrics, thereby achieving consistent gains over standard supervised fine-tuning. Additionally, we release the ScannedRxn benchmark, comprising scanned historical reaction diagrams with real-world artifacts, to rigorously assess model robustness and out-of-distribution ability. Our contributions advance the accuracy and generalization of VLM-based reaction diagram parsing. We will release data, models, and code on GitHub.

Hao Zheng, Shunzhi Yang, Zhuoxin He et al.

Pre-trained Vision-Language Models (VLMs) such as CLIP have shown excellent generalization abilities. However, adapting these large-scale models to downstream tasks while preserving their generalization capabilities remains challenging. Although prompt learning methods have shown promise, they suffer from two fundamental bottlenecks that limit generalization: (a) modality isolation, and (b) hierarchical semantic decay. To address these limitations, we propose HiCroPL, a Hierarchical Cross-modal Prompt Learning framework that establishes bidirectional knowledge flow between text and vision modalities, enabling them to refine their semantics mutually. HiCroPL routes knowledge flows by leveraging the complementary strengths of text and vision. In early layers, text prompts inject relatively clear semantics into visual prompts through a hierarchical knowledge mapper, enhancing the representation of low-level visual semantics. In later layers, visual prompts encoding specific task-relevant objects flow back to refine text prompts, enabling deeper alignment. Crucially, our hierarchical knowledge mapper allows representations at multi-scales to be fused, ensuring that deeper representations retain transferable shallow semantics thereby enhancing generalization. We further introduce a lightweight layer-specific knowledge proxy to enable efficient cross-modal interactions. Extensive evaluations across four tasks demonstrate HiCroPL's superior performance, achieving state-of-the-art results on 11 benchmarks with significant improvements. Code is available at: https://github.com/zzeoZheng/HiCroPL.

Yifei Sun, Zhanghao Chen, Hao Zheng et al.

Lung diseases represent a significant global health challenge, with Chest X-Ray (CXR) being a key diagnostic tool due to its accessibility and affordability. Nonetheless, the detection of pulmonary lesions is often hindered by overlapping bone structures in CXR images, leading to potential misdiagnoses. To address this issue, we develop an end-to-end framework called BS-LDM, designed to effectively suppress bone in high-resolution CXR images. This framework is based on conditional latent diffusion models and incorporates a multi-level hybrid loss-constrained vector-quantized generative adversarial network which is crafted for perceptual compression, ensuring the preservation of details. To further enhance the framework's performance, we utilize offset noise in the forward process, and a temporal adaptive thresholding strategy in the reverse process. These additions help minimize discrepancies in generating low-frequency information of soft tissue images. Additionally, we have compiled a high-quality bone suppression dataset named SZCH-X-Rays. This dataset includes 818 pairs of high-resolution CXR and soft tissue images collected from our partner hospital. Moreover, we processed 241 data pairs from the JSRT dataset into negative images, which are more commonly used in clinical practice. Our comprehensive experiments and downstream evaluations reveal that BS-LDM excels in bone suppression, underscoring its clinical value. Our code is available at https://github.com/diaoquesang/BS-LDM.

Sazzadul Islam, Tasnim Tabassum, Hao Zheng

Generating synthesizable Verilog for large, hierarchical hardware designs remains a significant challenge for large language models (LLMs), which struggle to replicate the structured reasoning that human experts employ when translating complex specifications into RTL. When tasked with producing hierarchical Verilog, LLMs frequently lose context across modules, hallucinate interfaces, fabricate inter-module wiring, and fail to maintain structural coherence - failures that intensify as design complexity grows and specifications involve informal prose, figures, and tables that resist direct operationalization. To address these challenges, we present VeriGraphi, a framework that introduces a spec-anchored Knowledge Graph as the architectural substrate driving the RTL generation pipeline. VeriGraphi constructs a HDA, a structured knowledge graph that explicitly encodes module hierarchy, port-level interfaces, wiring semantics, and inter-module dependencies as first-class graph entities and relations. Built through iterative multi-agent analysis of the specification, this Knowledge Graph provides a deterministic, machine-checkable structural scaffold before code generation. Guided by the KG, a progressive coding module incrementally generates pseudo-code and synthesizable RTL while enforcing interface consistency and dependency correctness at each submodule stage. We evaluate VeriGraphi on a benchmark of three representative specification documents from the National Institute of Standards and Technology and their corresponding implementations, and we present a RV32I processor as a detailed case study to illustrate the full pipeline. The results demonstrate that VeriGraphi enables reliable hierarchical RTL generation with minimal human intervention for RISC-V, marking a significant milestone for LLM-generated hardware design while maintaining strong functional correctness.

Can Sun, Hao Zheng, Zhigang Hu et al.

The single domain generalization(SDG) based on meta-learning has emerged as an effective technique for solving the domain-shift problem. However, the inadequate match of data distribution between source and augmented domains and difficult separation of domain-invariant features from domain-related features make SDG model hard to achieve great generalization. Therefore, a novel meta-learning method based on domain enhancement and feature alignment (MetaDefa) is proposed to improve the model generalization performance. First, the background substitution and visual corruptions techniques are used to generate diverse and effective augmented domains. Then, the multi-channel feature alignment module based on class activation maps and class agnostic activation maps is designed to effectively extract adequate transferability knowledge. In this module, domain-invariant features can be fully explored by focusing on similar target regions between source and augmented domains feature space and suppressing the feature representation of non-similar target regions. Extensive experiments on two publicly available datasets show that MetaDefa has significant generalization performance advantages in unknown multiple target domains.

Zixun Huang, Hao Zheng

In this paper, we propose a machine learning-based approach to address the lack of ability for designers to optimize urban land use planning from the perspective of vehicle travel demand. Research shows that our computational model can help designers quickly obtain feedback on the vehicle travel demand, which includes its total amount and temporal distribution based on the urban function distribution designed by the designers. It also assists in design optimization and evaluation of the urban function distribution from the perspective of vehicle travel. We obtain the city function distribution information and vehicle hours traveled (VHT) information by collecting the city point-of-interest (POI) data and online vehicle data. The artificial neural networks (ANNs) with the best performance in prediction are selected. By using data sets collected in different regions for mutual prediction and remapping the predictions onto a map for visualization, we evaluate the extent to which the computational model sees use across regions in an attempt to reduce the workload of future urban researchers. Finally, we demonstrate the application of the computational model to help designers obtain feedback on vehicle travel demand in the built environment and combine it with genetic algorithms to optimize the current state of the urban environment to provide recommendations to designers.

Jiahe Song, Chuang Wang, Bowen Jiang et al.

Large-scale chemical reaction datasets are crucial for AI research in chemistry. However, existing chemical reaction data often exist as images within papers, making them not machine-readable and unusable for training machine learning models. In response to this challenge, we propose the RxnCaption framework for the task of chemical Reaction Diagram Parsing (RxnDP). Our framework reformulates the traditional coordinate prediction driven parsing process into an image captioning problem, which Large Vision-Language Models (LVLMs) handle naturally. We introduce a strategy termed "BBox and Index as Visual Prompt" (BIVP), which uses our state-of-the-art molecular detector, MolYOLO, to pre-draw molecular bounding boxes and indices directly onto the input image. This turns the downstream parsing into a natural-language description problem. Extensive experiments show that the BIVP strategy significantly improves structural extraction quality while simplifying model design. We further construct the RxnCaption-11k dataset, an order of magnitude larger than prior real-world literature benchmarks, with a balanced test subset across four layout archetypes. Experiments demonstrate that RxnCaption-VL achieves state-of-the-art performance on multiple metrics. We believe our method, dataset, and models will advance structured information extraction from chemical literature and catalyze broader AI applications in chemistry. We will release data, models, and code on GitHub.

Junlai Qiu, Yunzhu Chen, Hao Zheng et al.

Diabetic retinopathy (DR) is a leading cause of vision loss among middle-aged and elderly people, which significantly impacts their daily lives and mental health. To improve the efficiency of clinical screening and enable the early detection of DR, a variety of automated DR diagnosis systems have been recently established based on convolutional neural network (CNN) or vision Transformer (ViT). However, due to the own shortages of CNN / ViT, the performance of existing methods using single-type backbone has reached a bottleneck. One potential way for the further improvements is integrating different kinds of backbones, which can fully leverage the respective strengths of them (\emph{i.e.,} the local feature extraction capability of CNN and the global feature capturing ability of ViT). To this end, we propose a novel paradigm to effectively fuse the features extracted by different backbones based on the theory of evidence. Specifically, the proposed evidential fusion paradigm transforms the features from different backbones into supporting evidences via a set of deep evidential networks. With the supporting evidences, the aggregated opinion can be accordingly formed, which can be used to adaptively tune the fusion pattern between different backbones and accordingly boost the performance of our hybrid model. We evaluated our method on two publicly available DR grading datasets. The experimental results demonstrate that our hybrid model not only improves the accuracy of DR grading, compared to the state-of-the-art frameworks, but also provides the excellent interpretability for feature fusion and decision-making.

Bardia Nadimi, Hao Zheng

Understanding communication behavior in modern system-on-chip (SoC) designs is critical for functional verification, performance analysis, and post-silicon debugging. Communication traces capture message exchanges among system components and provide valuable insights into system behavior. However, deriving concise communication specifications from such traces remains challenging due to interleaved instances of communication flows, and ambiguous causal relationships among messages. Existing mining approaches often struggle with scalability and ambiguity when traces contain complex interleaving of message patterns across multiple components. These conditions often lead to an explosion in the number of candidate flows and inaccurate extraction of communication behaviors. This paper presents AutoFlows++, a design-architecture-guided hierarchical framework for mining message flows from communication traces of complex SoC designs. AutoFlows++ operates in two stages: local mining followed by global mining. In the local mining stage, simple communication patterns are extracted from traces observed at individual communication interfaces between components. In the global mining stage, these local patterns are composed to identify higher-level message flows that characterize communication behavior across multiple components. Experimental results on both synthetic traces and traces generated from SoC models in GEM5 demonstrate that AutoFlows++ significantly improves flow extraction accuracy compared with prior approaches, highlighting its effectiveness for practical SoC validation tasks.

Xinyi Zheng, Yunze Liu, Chi-Hao Wu et al.

We present SpatialMem, a memory-centric system that unifies 3D geometry, semantics, and language into a single, queryable representation. Starting from casually captured egocentric RGB video, SpatialMem reconstructs metrically scaled indoor environments, detects structural 3D anchors (walls, doors, windows) as the first-layer scaffold, and populates a hierarchical memory with open-vocabulary object nodes -- linking evidence patches, visual embeddings, and two-layer textual descriptions to 3D coordinates -- for compact storage and fast retrieval. This design enables interpretable reasoning over spatial relations (e.g., distance, direction, visibility) and supports downstream tasks such as language-guided navigation and object retrieval without specialized sensors. Experiments across three real-life indoor scenes demonstrate that SpatialMem maintains strong anchor-description-level navigation completion and hierarchical retrieval accuracy under increasing clutter and occlusion, offering an efficient and extensible framework for embodied spatial intelligence.

Bardia Nadimi, Khashayar Filom, Deming Chen et al.

With the rapid advancement of Large Language Models (LLMs), there is growing interest in applying them to hardware design and verification. Among these stages, design verification remains the most time-consuming and resource-intensive phase, where generating effective stimuli for the design under test (DUT) is both critical and labor-intensive. We present {\it TB or not TB}, a framework for automated stimulus generation using LLMs fine-tuned through Coverage-Driven Direct Preference Optimization (CD-DPO). To enable preference-based training, we introduce PairaNet, a dataset derived from PyraNet that pairs high- and low-quality testbenches labeled using simulation-derived coverage metrics. The proposed CD-DPO method integrates quantitative coverage feedback directly into the optimization objective, guiding the model toward generating stimuli that maximize verification coverage. Experiments on the CVDP CID12 benchmark show that {\it TB or not TB} outperforms both open-source and commercial baselines, achieving up to 77.27\% improvement in code coverage, demonstrating the effectiveness of Coverage-driven preference optimization for LLM-based hardware verification.

Xuan Gong, Hanbo Huang, Hao Zheng et al.

Long chain-of-thought (CoT) reasoning improves large vision--language models, but visual information often fades during generation, limiting long-horizon multimodal reasoning. Existing methods either re-inject vision at inference or train policies for stronger grounding, but where to intervene relies on perception heuristics rather than principled gain analysis, and how local visual influence propagates remains implicit. We study this problem from an information-theoretic standpoint and derive a lower bound on the downstream visual gain of a one-step intervention, which suggests two factors: local branching room (token entropy) and downstream visual propagation potential (suffix divergence from a vision-marginalized reference). Guided by this analysis, we propose reflection-anchor policy optimization (RAPO), a GRPO-based policy optimization method that selects high-entropy reflection anchors and optimizes a chain-masked finite-window KL surrogate for downstream visual dependence. Experiments on reasoning-intensive and general-domain benchmarks show that RAPO delivers substantial gains over strong baselines across multiple LVLM backbones. Mechanism analyses further indicate that reflection anchors are enriched for visually sensitive decision points and that RAPO increases contrastive visual-dependence signals along generated trajectories.

Hao Zheng, Zirui Pang, Ling li et al.

Advances in Multimodal Large Language Models (MLLMs) intensify concerns about data privacy, making Machine Unlearning (MU), the selective removal of learned information, a critical necessity. However, existing MU benchmarks for MLLMs are limited by a lack of image diversity, potential inaccuracies, and insufficient evaluation scenarios, which fail to capture the complexity of real-world applications. To facilitate the development of MLLMs unlearning and alleviate the aforementioned limitations, we introduce OFFSIDE, a novel benchmark for evaluating misinformation unlearning in MLLMs based on football transfer rumors. This manually curated dataset contains 15.68K records for 80 players, providing a comprehensive framework with four test sets to assess forgetting efficacy, generalization, utility, and robustness. OFFSIDE supports advanced settings like selective unlearning and corrective relearning, and crucially, unimodal unlearning (forgetting only text data). Our extensive evaluation of multiple baselines reveals key findings: (1) Unimodal methods (erasing text-based knowledge) fail on multimodal rumors; (2) Unlearning efficacy is largely driven by catastrophic forgetting; (3) All methods struggle with "visual rumors" (rumors appear in the image); (4) The unlearned rumors can be easily recovered and (5) All methods are vulnerable to prompt attacks. These results expose significant vulnerabilities in current approaches, highlighting the need for more robust multimodal unlearning solutions. The code is available at \href{https://github.com/zh121800/OFFSIDE}{https://github.com/zh121800/OFFSIDE}.

Yifei Sun, Zhanghao Chen, Hao Zheng et al.

Chest X-Ray (CXR) imaging for pulmonary diagnosis raises significant challenges, primarily because bone structures can obscure critical details necessary for accurate diagnosis. Recent advances in deep learning, particularly with diffusion models, offer significant promise for effectively minimizing the visibility of bone structures in CXR images, thereby improving clarity and diagnostic accuracy. Nevertheless, existing diffusion-based methods for bone suppression in CXR imaging struggle to balance the complete suppression of bones with preserving local texture details. Additionally, their high computational demand and extended processing time hinder their practical use in clinical settings. To address these limitations, we introduce a Global-Local Latent Consistency Model (GL-LCM) architecture. This model combines lung segmentation, dual-path sampling, and global-local fusion, enabling fast high-resolution bone suppression in CXR images. To tackle potential boundary artifacts and detail blurring in local-path sampling, we further propose Local-Enhanced Guidance, which addresses these issues without additional training. Comprehensive experiments on a self-collected dataset SZCH-X-Rays, and the public dataset JSRT, reveal that our GL-LCM delivers superior bone suppression and remarkable computational efficiency, significantly outperforming several competitive methods. Our code is available at https://github.com/diaoquesang/GL-LCM.

Conghao Xiong, Hao Chen, Hao Zheng et al.

Survival analysis, as a challenging task, requires integrating Whole Slide Images (WSIs) and genomic data for comprehensive decision-making. There are two main challenges in this task: significant heterogeneity and complex inter- and intra-modal interactions between the two modalities. Previous approaches utilize co-attention methods, which fuse features from both modalities only once after separate encoding. However, these approaches are insufficient for modeling the complex task due to the heterogeneous nature between the modalities. To address these issues, we propose a Biased Progressive Encoding (BPE) paradigm, performing encoding and fusion simultaneously. This paradigm uses one modality as a reference when encoding the other. It enables deep fusion of the modalities through multiple alternating iterations, progressively reducing the cross-modal disparities and facilitating complementary interactions. Besides modality heterogeneity, survival analysis involves various biomarkers from WSIs, genomics, and their combinations. The critical biomarkers may exist in different modalities under individual variations, necessitating flexible adaptation of the models to specific scenarios. Therefore, we further propose a Mixture of Multimodal Experts (MoME) layer to dynamically selects tailored experts in each stage of the BPE paradigm. Experts incorporate reference information from another modality to varying degrees, enabling a balanced or biased focus on different modalities during the encoding process. Extensive experimental results demonstrate the superior performance of our method on various datasets, including TCGA-BLCA, TCGA-UCEC and TCGA-LUAD. Codes are available at https://github.com/BearCleverProud/MoME.

Runqi Wang, Hao Zheng, Xiaoyue Duan et al.

Pre-trained vision-language models have inspired much research on few-shot learning. However, with only a few training images, there exist two crucial problems: (1) the visual feature distributions are easily distracted by class-irrelevant information in images, and (2) the alignment between the visual and language feature distributions is difficult. To deal with the distraction problem, we propose a Selective Attack module, which consists of trainable adapters that generate spatial attention maps of images to guide the attacks on class-irrelevant image areas. By messing up these areas, the critical features are captured and the visual distributions of image features are calibrated. To better align the visual and language feature distributions that describe the same object class, we propose a cross-modal distribution alignment module, in which we introduce a vision-language prototype for each class to align the distributions, and adopt the Earth Mover's Distance (EMD) to optimize the prototypes. For efficient computation, the upper bound of EMD is derived. In addition, we propose an augmentation strategy to increase the diversity of the images and the text prompts, which can reduce overfitting to the few-shot training images. Extensive experiments on 11 datasets demonstrate that our method consistently outperforms prior arts in few-shot learning. The implementation code will be available at https://github.com/bhrqw/SADA.