arXiv subjects

Junjie Zhou

Publications and source records attributed to Junjie Zhou.

At least 37 records · Page 2Linked to original sources

arXiv2026

Advancing Cancer Prognosis with Hierarchical Fusion of Genomic, Proteomic and Pathology Imaging Data from a Systems Biology Perspective

To enhance the precision of cancer prognosis, recent research has increasingly focused on multimodal survival methods by integrating genomic data and histology images. However, current approaches overlook the fact that the proteome serves as an intermediate layer bridging genomic alterations and histopathological features while providing complementary biological information essential for survival prediction. This biological reality exposes another architectural limitation: existing integrative analysis studies fuse these heterogeneous data sources in a flat manner that fails to capture their inherent biological hierarchy. To address these limitations, we propose HFGPI, a hierarchical fusion framework that models the biological progression from genes to proteins to histology images from a systems biology perspective. Specifically, we introduce Molecular Tokenizer, a molecular encoding strategy that integrates identity embeddings with expression profiles to construct biologically informed representations for genes and proteins. We then develop Gene-Regulated Protein Fusion (GRPF), which employs graph-aware cross-attention with structure-preserving alignment to explicitly model gene-protein regulatory relationships and generate gene-regulated protein representations. Additionally, we propose Protein-Guided Hypergraph Learning (PGHL), which establishes associations between proteins and image patches, leveraging hypergraph convolution to capture higher-order protein-morphology relationships. The final features are progressively fused across hierarchical layers to achieve precise survival outcome prediction. Extensive experiments on five benchmark datasets demonstrate the superiority of HFGPI over state-of-the-art methods.

cs.CV↗

arXiv2026

Price Regulation and Network Spillovers

We study price regulation for a monopolist operating in networked markets with demand spillovers. Achieving efficiency requires price reductions proportional to consumers' Katz-Bonacich centralities, which generally cannot be implemented by commonly used price regulations. Moreover, these regulations become asymptotically welfare neutral as spillovers grow. Nevertheless, some price regulations may still benefit consumers. In particular, average-price regulation robustly increases consumer surplus. By contrast, banning price discrimination increases consumer surplus only when more central consumers have higher intrinsic willingness to pay.

econ.TH↗

arXiv2026

Beyond One-Thousandth Energy Resolution with an AlMn TES Detector

The superconducting Transition-Edge Sensor (TES) is a critical technology for next-generation X-ray spectrometers, known for its exceptional energy resolution. In the last decade, TESs based on AlMn alloy films have been extensively used in several cosmic microwave background (CMB) experiments. The advantages of simple fabrication process and easily tunable critical temperature make them an alternative to bilayer TESs. However, they have rarely been applied to X-ray detection until now. We developed an annular AlMn TES for X-ray detection and tested it in a dilution refrigerator with a Superconducting Quantum Interference Device (SQUID) amplifier, achieving an Full Width at Half Maximum (FWHM) of 12.1 +- 0.3 eV at 17.48 keV. To the best of our knowledge, this is the first demonstration of an AlMn TES achieving an energy resolution below 0.1%, highlighting its potential for high-resolution X-ray detection.

astro-ph.IM↗

arXiv2026

Human Identification at a Distance: Challenges, Methods and Results on the Competition HID 2025

Human identification at a distance (HID) is challenging because traditional biometric modalities such as face and fingerprints are often difficult to acquire in real-world scenarios. Gait recognition provides a practical alternative, as it can be captured reliably at a distance. To promote progress in gait recognition and provide a fair evaluation platform, the International Competition on Human Identification at a Distance (HID) has been organized annually since 2020. Since 2023, the competition has adopted the challenging SUSTech-Competition dataset, which features substantial variations in clothing, carried objects, and view angles. No dedicated training data are provided, requiring participants to train their models using external datasets. Each year, the competition applies a different random seed to generate distinct evaluation splits, which reduces the risk of overfitting and supports a fair assessment of cross-domain generalization. While HID 2023 and HID 2024 already used this dataset, HID 2025 explicitly examined whether algorithmic advances could surpass the accuracy limits observed previously. Despite the heightened difficulty, participants achieved further improvements, and the best-performing method reached 94.2% accuracy, setting a new benchmark on this dataset. We also analyze key technical trends and outline potential directions for future research in gait recognition.

cs.CV↗

arXiv2026

GenAgent: Scaling Text-to-Image Generation via Agentic Multimodal Reasoning

We introduce GenAgent, unifying visual understanding and generation through an agentic multimodal model. Unlike unified models that face expensive training costs and understanding-generation trade-offs, GenAgent decouples these capabilities through an agentic framework: understanding is handled by the multimodal model itself, while generation is achieved by treating image generation models as invokable tools. Crucially, unlike existing modular systems constrained by static pipelines, this design enables autonomous multi-turn interactions where the agent generates multimodal chains-of-thought encompassing reasoning, tool invocation, judgment, and reflection to iteratively refine outputs. We employ a two-stage training strategy: first, cold-start with supervised fine-tuning on high-quality tool invocation and reflection data to bootstrap agent behaviors; second, end-to-end agentic reinforcement learning combining pointwise rewards (final image quality) and pairwise rewards (reflection accuracy), with trajectory resampling for enhanced multi-turn exploration. GenAgent significantly boosts base generator(FLUX.1-dev) performance on GenEval++ (+23.6\%) and WISE (+14\%). Beyond performance gains, our framework demonstrates three key properties: 1) cross-tool generalization to generators with varying capabilities, 2) test-time scaling with consistent improvements across interaction rounds, and 3) task-adaptive reasoning that automatically adjusts to different tasks. Our code will be available at \href{https://github.com/deep-kaixun/GenAgent}{this url}.

cs.CV↗

arXiv2026

MomentSeeker: A Task-Oriented Benchmark For Long-Video Moment Retrieval

Accurately locating key moments within long videos is crucial for solving long video understanding (LVU) tasks. However, existing benchmarks are either severely limited in terms of video length and task diversity, or they focus solely on the end-to-end LVU performance, making them inappropriate for evaluating whether key moments can be accurately accessed. To address this challenge, we propose MomentSeeker, a novel benchmark for long-video moment retrieval (LMVR), distinguished by the following features. First, it is created based on long and diverse videos, averaging over 1200 seconds in duration and collected from various domains, e.g., movie, anomaly, egocentric, and sports. Second, it covers a variety of real-world scenarios in three levels: global-level, event-level, object-level, covering common tasks like action recognition, object localization, and causal reasoning, etc. Third, it incorporates rich forms of queries, including text-only queries, image-conditioned queries, and video-conditioned queries. On top of MomentSeeker, we conduct comprehensive experiments for both generation-based approaches (directly using MLLMs) and retrieval-based approaches (leveraging video retrievers). Our results reveal the significant challenges in long-video moment retrieval in terms of accuracy and efficiency, despite improvements from the latest long-video MLLMs and task-specific fine-tuning. We have publicly released MomentSeeker(https://yhy-2000.github.io/MomentSeeker/) to facilitate future research in this area.

cs.CV↗

arXiv2025

Sequential Network Design

We study dynamic network formation from a centralized perspective. In each period, the social planner builds a single link to connect previously unlinked pairs. The social planner is forward-looking, with instantaneous utility monotonic in the aggregate number of walks of various lengths. We show that, forming a nested split graph at each period is optimal, regardless of the discount function. When the social planner is sufficiently myopic, it is optimal to form a quasi-complete graph at each period, which is unique up to permutation. This finding provides a micro-foundation for the quasi-complete graph, as it is formed under a greedy policy. We also investigate the robustness of these findings under non-linear best response functions and weighted networks.

econ.TH↗

arXiv2025

Welfare and Distributional Effects of Joint Intervention in Networks

We study the optimal joint intervention of a planner who can influence both the standalone marginal utilities of agents in a network and the weights of the links connecting them. The welfare-maximizing intervention displays two key features. First, when the planner's budget is moderate (yielding interior solutions), the optimal change in link weight between any pair of agents is proportional to the product of their eigen-centralities. Second, when the budget is sufficiently large, the optimal network converges to a simple structure: a complete network under strategic complements, or a complete balanced bipartite network under strategic substitutes. We show that welfare effects are governed by the principal eigenvalue of the network, while distributional outcomes are driven by the dispersion of the corresponding eigen-centralities. Comparing joint interventions to single interventions targeting only standalone marginal utilities, we find that joint interventions consistently generate higher aggregate welfare, but may also increase inequality, revealing a potential trade-off between efficiency and equity.

econ.TH↗

arXiv2025

VideoExplorer: Think With Videos For Agentic Long-Video Understanding

Long-video understanding~(LVU) is a challenging problem in computer vision. Existing methods either downsample frames for single-pass reasoning, sacrificing fine-grained details, or depend on textual reasoning over task-agnostic representations, hindering task-specific perception and exploration. In this paper, we propose VideoExplorer, a framework grounded in the principle of ``thinking with video'', which naturally intertwines planning, temporal grounding, and scalable perception into a coherent reasoning process. Rather than reasoning over a static context, VideoExplorer iteratively formulates sub-questions, locates relevant moments, and performs task-oriented, temporally scalable video understanding until reaching the final answer, enabling faithful, efficient, and interpretable reasoning. To address the lack of LVU training resources, we construct a long-video reasoning dataset using difficulty-adaptive sampling to ensure high-quality trajectories on complex tasks. Building on this dataset, we design a two-stage training pipeline: supervised trajectory initialization followed by trajectory-level preference optimization, encouraging adaptive temporal grounding and iterative information integration guided by downstream rewards. Extensive evaluations on popular long-video understanding and reasoning benchmarks demonstrate VideoExplorer's significant advantage over existing baselines, highlighting its robustness, adaptability, and efficiency. Our code is made publicly available in this repository(https://github.com/yhy-2000/VideoDeepResearch).

cs.CV↗

arXiv2025

Interconnected Contests

We study a two-player model of conflict with multiple battlefields -- the novel element is that each of the players has their own network of spillovers so that resources allocated to one battle can be utilized in winning neighboring battles. There exists a unique equilibrium in which the relative probability of a player winning a battle is the product of the ratio of the centrality of the battlefield in the two respective competing networks and the ratio of the relative cost of efforts of the two players. We study the design of networks and characterize networks that maximize total efforts and maximize total utility. Finally, we characterize the equilibrium of a game in which players choose both networks and efforts in the battles.

econ.TH↗

arXiv2025

Fabrication and Characterization of X-ray TES Detectors Based on Annular AlMn Alloy Films

AlMn alloy flms are widely fabricated into superconducting transition edge sensors (TESs) for the detection of cosmic microwave background radiation. However, the application in X-ray or gamma-ray detection based on AlMn TES is rarely reported. In this study, X-ray TES detectors based on unique annular AlMn flms are devel-oped. The fabrication processes of TES detectors are introduced in detail. The char-acteristics of three TES samples are evaluated in a dilution refrigerator. The results demonstrate that the I-V characteristics of the three annular TES detectors are highly consistent. The TES detector with the smallest absorber achieved the best energy resolution of 11.0 eV @ 5.9 keV, which is inferior to the theoretical value. The dis-crepancy is mainly attributed to the larger readout electronics noise than expected.

physics.ins-det↗

arXiv2025

MAPLE: Multi-scale Attribute-enhanced Prompt Learning for Few-shot Whole Slide Image Classification

Prompt learning has emerged as a promising paradigm for adapting pre-trained vision-language models (VLMs) to few-shot whole slide image (WSI) classification by aligning visual features with textual representations, thereby reducing annotation cost and enhancing model generalization. Nevertheless, existing methods typically rely on slide-level prompts and fail to capture the subtype-specific phenotypic variations of histological entities (\emph{e.g.,} nuclei, glands) that are critical for cancer diagnosis. To address this gap, we propose Multi-scale Attribute-enhanced Prompt Learning (\textbf{MAPLE}), a hierarchical framework for few-shot WSI classification that jointly integrates multi-scale visual semantics and performs prediction at both the entity and slide levels. Specifically, we first leverage large language models (LLMs) to generate entity-level prompts that can help identify multi-scale histological entities and their phenotypic attributes, as well as slide-level prompts to capture global visual descriptions. Then, an entity-guided cross-attention module is proposed to generate entity-level features, followed by aligning with their corresponding subtype-specific attributes for fine-grained entity-level prediction. To enrich entity representations, we further develop a cross-scale entity graph learning module that can update these representations by capturing their semantic correlations within and across scales. The refined representations are then aggregated into a slide-level representation and aligned with the corresponding prompts for slide-level prediction. Finally, we combine both entity-level and slide-level outputs to produce the final prediction results. Results on three cancer cohorts confirm the effectiveness of our approach in addressing few-shot pathology diagnosis tasks.

cs.CV↗

arXiv2025

MR$^2$-Bench: Going Beyond Matching to Reasoning in Multimodal Retrieval

Multimodal retrieval is becoming a crucial component of modern AI applications, yet its evaluation lags behind the demands of more realistic and challenging scenarios. Existing benchmarks primarily probe surface-level semantic correspondence (e.g., object-text matching) while failing to assess the deeper reasoning required to capture complex relationships between visual and textual information. To address this gap, we introduce MR$^2$-Bench, a reasoning-intensive benchmark for multimodal retrieval. MR$^2$-Bench presents the following critical values: 1) all tasks are reasoning-driven, going beyond shallow matching to effectively assess models' capacity for logical, spatial, and causal inference; 2) it features diverse multimodal data, such as natural images, diagrams, and visual puzzles, enabling comprehensive evaluation across content types; 3) it supports complex queries and documents containing multiple images and covers diverse retrieval scenarios, more accurately reflecting real-world applications. Our benchmark contains 1,309 curated queries, derived either from manual collection and annotation or from selective consolidation of public datasets. Despite achieving strong results on existing benchmarks, current state-of-the-art models still struggle on MR$^2$-Bench: for example, the leading Seed1.6-Embedding model attains a Recall@1 of 77.78 on MMEB, but only 9.91 on MR$^2$-Bench. This substantial performance gap highlights both the increased challenge posed by our benchmark and the pressing need for further advances in reasoning-intensive multimodal retrieval. The dataset and evaluation code will be made publicly available at https://github.com/VectorSpaceLab/MR2-Bench.

cs.IR↗

arXiv2025

CGTGait: Collaborative Graph and Transformer for Gait Emotion Recognition

Skeleton-based gait emotion recognition has received significant attention due to its wide-ranging applications. However, existing methods primarily focus on extracting spatial and local temporal motion information, failing to capture long-range temporal representations. In this paper, we propose \textbf{CGTGait}, a novel framework that collaboratively integrates graph convolution and transformers to extract discriminative spatiotemporal features for gait emotion recognition. Specifically, CGTGait consists of multiple CGT blocks, where each block employs graph convolution to capture frame-level spatial topology and the transformer to model global temporal dependencies. Additionally, we introduce a Bidirectional Cross-Stream Fusion (BCSF) module to effectively aggregate posture and motion spatiotemporal features, facilitating the exchange of complementary information between the two streams. We evaluate our method on two widely used datasets, Emotion-Gait and ELMD, demonstrating that our CGTGait achieves state-of-the-art or at least competitive performance while reducing computational complexity by approximately \textbf{82.2\%} (only requiring 0.34G FLOPs) during testing. Code is available at \small{https://github.com/githubzjj1/CGTGait.}

cs.CV↗

arXiv2025

"One defect, one potential" strategy for accurate machine learning prediction of defect phonons

Atomic vibrations play a critical role in phonon-assisted electron transitions at defects in solids. However, accurate phonon calculations in defect systems are often hindered by the high computational cost of large-supercell first-principles calculations. Recently, foundation models, such as universal machine learning interatomic potentials (MLIPs), emerge as a promising alternative for rapid phonon calculations, but the quantitatively low accuracy restricts its fundamental applicability for high-level defect phonon calculations, such as nonradiative carrier capture rates. In this paper, we propose a "one defect, one potential" strategy in which an MLIP is trained on a limited set of perturbed supercells. We demonstrate that this strategy yields phonons with accuracy comparable to density functional theory (DFT), regardless of the supercell size. The predicted accuracy of defect phonons is validated by phonon frequencies, Huang-Rhys factors, and phonon dispersions. Further calculations of photoluminescence (PL) spectra and nonradiative capture rates based on this defect-specific model also show good agreements with DFT results, meanwhile reducing the computational expenses by more than an order of magnitude. Our approach provides a practical pathway for studying defect phonons in 10$^4$-atom large supercell with high accuracy and efficiency.

cond-mat.mtrl-sci↗

arXiv2025

Influence and Connectivity in Networks: A Generating Function Approach

Many widely used network centralities are based on counting walks that meet specific criteria. This paper introduces a systematic framework for walk enumeration using generating functions. We introduce a first-passage decomposition that uniquely divides any walk passing through specified nodes or links into two components: a first-reaching walk and a subsequent walk. This decomposition yields a system of interconnected equations that relate three disjoint categories of walks: unrestricted walks, walks that avoid specific elements, and walks that pass through designated sets. The framework offers a range of applications, including evaluating the effects of structural interventions, such as node or link modifications, on network walks, generalizing target centrality to multi-receiver scenarios in information networks, and comparing different strategies for adding links.

econ.TH↗

arXiv2025

Quantization without Tears

Deep neural networks, while achieving remarkable success across diverse tasks, demand significant resources, including computation, GPU memory, bandwidth, storage, and energy. Network quantization, as a standard compression and acceleration technique, reduces storage costs and enables potential inference acceleration by discretizing network weights and activations into a finite set of integer values. However, current quantization methods are often complex and sensitive, requiring extensive task-specific hyperparameters, where even a single misconfiguration can impair model performance, limiting generality across different models and tasks. In this paper, we propose Quantization without Tears (QwT), a method that simultaneously achieves quantization speed, accuracy, simplicity, and generality. The key insight of QwT is to incorporate a lightweight additional structure into the quantized network to mitigate information loss during quantization. This structure consists solely of a small set of linear layers, keeping the method simple and efficient. More importantly, it provides a closed-form solution, allowing us to improve accuracy effortlessly under 2 minutes. Extensive experiments across various vision, language, and multimodal tasks demonstrate that QwT is both highly effective and versatile. In fact, our approach offers a robust solution for network quantization that combines simplicity, accuracy, and adaptability, which provides new insights for the design of novel quantization paradigms. The code is publicly available at https://github.com/wujx2001/QwT

cs.CV↗

arXiv2025

PathCoT: Chain-of-Thought Prompting for Zero-shot Pathology Visual Reasoning

With the development of generative artificial intelligence and instruction tuning techniques, multimodal large language models (MLLMs) have made impressive progress on general reasoning tasks. Benefiting from the chain-of-thought (CoT) methodology, MLLMs can solve the visual reasoning problem step-by-step. However, existing MLLMs still face significant challenges when applied to pathology visual reasoning tasks: (1) LLMs often underperforms because they lack domain-specific information, which can lead to model hallucinations. (2) The additional reasoning steps in CoT may introduce errors, leading to the divergence of answers. To address these limitations, we propose PathCoT, a novel zero-shot CoT prompting method which integrates the pathology expert-knowledge into the reasoning process of MLLMs and incorporates self-evaluation to mitigate divergence of answers. Specifically, PathCoT guides the MLLM with prior knowledge to perform as pathology experts, and provides comprehensive analysis of the image with their domain-specific knowledge. By incorporating the experts' knowledge, PathCoT can obtain the answers with CoT reasoning. Furthermore, PathCoT incorporates a self-evaluation step that assesses both the results generated directly by MLLMs and those derived through CoT, finally determining the reliable answer. The experimental results on the PathMMU dataset demonstrate the effectiveness of our method on pathology visual understanding and reasoning.

cs.LG↗