Searcharxiv⌕ Search

arXiv subjects

Jun Yin

Publications and source records attributed to Jun Yin.

At least 37 records · Page 2Linked to original sources
arXiv2026

GRACE: Boosting Video MLLMs with Grounded Action-Centric Evidence for Viewer Sentiment Prediction

Viewer sentiment prediction in video advertisements aims to infer the latent affective response evoked in the audience. To bridge the gap between what is shown and what is felt, models must deduce hidden viewer emotions from explicit visual narratives, concrete character-object interactions, and visible textual cues. However, standard Multimodal Large Language Models (MLLMs) typically rely on holistic frame representations, which leave these fine-grained, affect-relevant events implicit and complicate precise emotional reasoning. To address this, we propose a grounded action-centric evidence augmentation framework that enhances video MLLMs' clue extraction and comprehension by introducing explicit event structure and localized visual evidence. Our method extracts temporally ordered subject-verb-object (SVO) triplets and auxiliary visible textual cues from action-centric video descriptions, grounds subject and object entities as visual entity crops, and then enables the MLLM to perform clue-enhanced emotional reasoning based on these extracted structured clues. In this way, action triplets specify "what happens", while grounded visual entity crops anchor "who or what participates in each event" to concrete visual evidence. Experiments on the Pitts dataset show consistent improvements over Qwen2.5-VL and Qwen3-VL baselines. Ablation studies, cross-dataset evaluation on AdsQA, and transfer experiments on an emotion-focused TVQA subset further support the effectiveness and generalization of our approach.

cs.CV↗
arXiv2026

JailbreakOPT: Tool-Assisted Iterative Jailbreak Prompt Optimization

Jailbreak attacks expose persistent safety weaknesses in large language models (LLMs), but existing stateless single-turn methods face a trade-off: hand-crafted prompts are expressive but static, while iterative prompt optimization can adapt but often relies on low-level mutations that require many target queries. We propose JailbreakOPT, a tool-assisted framework for improving iterative single-turn jailbreak prompt optimization. JailbreakOPT organizes diverse atomic jailbreak prompts into an attack tool library and composes them through a unified intra-episode optimization abstraction to generate stronger standalone attack prompts. To reuse experience across attack episodes, JailbreakOPT further frames tool selection as a contextual bandit problem and applies contextual Thompson sampling to guide exploration and exploitation based on past outcomes. Experiments across multiple target LLMs and attack goals show that JailbreakOPT improves attack success rate (ASR) while reducing the number of attacks until success (No.A) compared with atomic single-turn attacks and existing iterative optimization baselines. This paper may contain offensive or harmful content.

cs.CR↗
arXiv2026

Detect in Any Scene: An Agentic Framework for Object Detection with Experience-Aware Reasoning

Object detection in real-world scenarios remains challenging due to diverse image degradations and heterogeneous object distributions, which significantly hinder the generalization of existing detectors. Conventional approaches, including scene-specific representation learning and end-to-end pipeline design, are inherently limited by their reliance on predefined conditions and lack adaptability to dynamic environments. In this paper, we propose DetAS, an agentic detection framework that formulates object detection as a dynamic decision process. Instead of relying on static pipelines, DetAS leverages a Multimodal Large Language Model (MLLM) as a central agent to adaptively compose detection workflows by selecting from a toolbox of restoration modules and specialized detectors. Specifically, DetAS consists of two key components: Self-Adaptive Image Restoration, which dynamically determines whether and how to enhance images for downstream detection, and Multi-Expertise Detection, which integrates multiple domain-specialized detectors and resolves their predictions through instance-level reasoning. To further improve decision quality under fine-grained conditions, we introduce Self-Evolving Experience Harvesting and extend the framework to DetAS-X, which accumulates node-level decision experience from a small set of annotated data and enables experience-aware reasoning during inference. This mechanism allows the system to progressively refine its decision policy and adapt to diverse real-world scenarios. Extensive experiments on six challenging benchmarks demonstrate that DetAS-X significantly outperforms existing MLLM-based detectors, achieving an average improvement of 28.36% in F1 score, with up to 37.01% gain on DarkFace. These results demonstrate the promise of agentic detection and establish a solid foundation for its application in complex and dynamic environments.

cs.CV↗
arXiv2026

Rel-MOSS: Towards Imbalanced Relational Deep Learning on Relational Databases

In recent advances, to enable a fully data-driven learning paradigm on relational databases (RDB), relational deep learning (RDL) is proposed to structure the RDB as a heterogeneous entity graph and adopt the graph neural network (GNN) as the predictive model. However, existing RDL methods neglect the imbalance problem of relational data in RDBs and risk under-representing the minority entities, leading to an unusable model in practice. In this work, we investigate, for the first time, class imbalance problem in RDB entity classification and design the relation-centric minority synthetic over-sampling GNN (Rel-MOSS), in order to fill a critical void in the current literature. Specifically, to mitigate the issue of minority-related information being submerged by majority counterparts, we design the relation-wise gating controller to modulate neighborhood messages from each individual relation type. Based on the relational-gated representations, we further propose the relation-guided minority synthesizer for over-sampling, which integrates the entity relational signatures to maintain relational consistency. Extensive experiments on 12 entity classification datasets provide compelling evidence for the superiority of Rel-MOSS, yielding an average improvement of up to 2.46% and 4.00% in terms of Balanced Accuracy and G-Mean, compared with SOTA RDL methods and classic methods for handling class imbalance.

cs.AI↗
arXiv2026

Echoes in Filter Bubble: Diagnosing and Curing Popularity Bias in Generative Recommenders

Recently, Generative Recommenders (GRs), characterized by a unified end-to-end framework, have exhibited astonishing potential in transforming the recommendation paradigm. Despite their effectiveness, we recognize that GRs are still susceptible to the long-standing issue of popularity bias that has pervaded the recommendation community. Although a few studies have attempted to extend traditional debiasing methods to GRs, their effectiveness is marginal, and the fundamental reason why GRs suffer from popularity bias remains under-explored. To bridge this gap, this study focuses on two core aspects in GRs: the optimization of generative framework and the item tokenization based on semantic index. Based on theoretical analyses, we identify that the severe popularity bias emerges from the confluence of a token-level optimization flaw and the undifferentiated property of item tokenization. Accordingly, this study develops a novel generative recommender system, called Ghost, by designing the asymmetric unlikelihood optimization and the skeleton-founded tokenization. Extensive empirical evaluations across three datasets, alongside multiple SOTA baselines, reveal that Ghost substantially alleviates popularity bias and promotes fairer recommendations, while incurring slight degradation to the overall recommendation utility.

cs.IR↗
arXiv2026

Beyond the Aggregation Dilemma: Prior-Retaining Decoupled Learning for Multimodal Graphs

Multimodal Attributed Graph Learning (MAGL) integrates intrinsic node attributes with structural topology via graph aggregation. However, as pretrained encoders evolve into Large Foundation Models (LFMs), the landscape of MAGL fundamentally shifts: under high-confidence LFM priors, mandatory aggregation introduces topological noise that overwhelms discriminative signals, triggering a counter-intuitive performance inversion where sophisticated MAGL architectures underperform simple topology-agnostic MLPs. Through systematic empirical and theoretical analysis, we identify that this inversion stems from a fundamental aggregation dilemma characterized by two concurrent pathologies: (1) Representational Pathology (SNR Degradation) - mandatory aggregation dilutes robust intrinsic features with topological noise, causing the noise penalty to outweigh its collaborative benefit; and (2) Optimization Pathology (Gradient Starvation) - topological aggregation attenuates gradient flow, while a shared task loss causes dominant modalities to prematurely suppress weaker ones. To resolve this dilemma, we propose SUPRA (Shared-Unique Prior-Retaining Architecture), a decoupled dual-pathway paradigm. SUPRA processes modality-specific features through topology-agnostic MLPs while capturing structural synergy via a lightweight shared GNN, with auxiliary deep supervision counteracting gradient starvation. Extensive evaluations demonstrate that SUPRA achieves state-of-the-art performance while requiring 3.5x lower peak GPU memory and up to 4.4x faster training time than Multimodal Graph Transformers.

cs.LG↗
arXiv2026

Power Reinforcement Post-Training of Text-to-Image Models with Super-Linear Advantage Shaping

Recently, post-training methods based on reinforcement learning, with a particular focus on Group Relative Policy Optimization (GRPO), have emerged as the robust paradigm for further advancement of text-to-image (T2I) models. However, these methods are often prone to reward hacking, wherein models exploit biases in imperfect reward functions rather than yielding genuine performance gains. In this work, we identify that normalization could lead to miscalibration and directly removing the prompt-level standard deviation term yields an optimal policy ascent direction that is linear in the advantage but still limits the separation of genuine signals from noise. To mitigate the above issues, we propose Super-Linear Advantage Shaping (SLAS) by revisiting the functional update from an information geometry perspective. By extending the Fisher-Rao information metric with advantage-dependent weighting, SLAS introduces a non-linear geometric structure that reshapes the local policy space. This design relaxes constraints along high-advantage directions to amplify informative updates, while tightening those in low-advantage regions to suppress illusory gradients. In addition, batch-level normalization is applied to stabilize training under varying reward scales. Extensive evaluations demonstrate that SLAS consistently surpasses the DanceGRPO baseline across multiple backbones and benchmarks. In particular, it yields faster training dynamics, improved out-of-domain performance on GenEval and UniGenBench++, and enhanced robustness to model scaling, while mitigating reward hacking and preserving semantic and compositional fidelity in generations.

cs.CV↗
arXiv2026

Chiral superfluorescence from perovskite superlattices at room temperature

Superfluorescence (SF) is the collective emission of intense, coherent light from an interacting ensemble of quantum emitters1-4. While SF has been observed in several solid-state materials5-8, the spontaneous generation of circularly polarized SF from chiral materials (chiral SF) has not been realized9,10. Here, we report the first observation of chiral SF originating from edge states in large-area (>100 um * 100 um) vertically aligned chiral perovskite superlattices at room-temperature. Theoretical quantum optics calculations describe the transition from initially unpolarized, incoherent spontaneous emission to a coherent chiral SF state, quantitatively reproducing both the experimentally observed generation of circular polarization (up to ~14%) and its reversal in sign with opposite material handedness. Moreover, we show that both the intensity and the degree of circular polarization of chiral SF can be modulated by a weak magnetic field, enabling precise control over solid-state quantum light emission at room temperature. Our findings demonstrate an interplay between chirality and many-body quantum coherence, thereby revealing promising new directions for chirality-controlled quantum-optical applications.

physics.optics↗
arXiv2026

Beyond Mass and Multiscale Environments: What Shapes Low Surface Brightness Galaxies? Evidence from MaNGA

The origin of low surface brightness (LSB) galaxies remains a key open question in galaxy formation, reflecting the balance internal mechanisms and environmental influence. Using MaNGA integral-field spectroscopy, we investigate whether LSB and high surface brightness (HSB) galaxies of comparable stellar mass ($9 < \log M_\ast < 10$) occupy distinct environments or differ primarily through internal evolution. Our late-type sample comprises 113 central and 29 satellite LSB galaxies, and 374 central and 142 satellite HSB galaxies. We characterize environments on scales from 100 kpc to 10 Mpc, analyzing radial profiles of stellar mass surface density ($Σ_\ast$), star formation activity, and gas-phase metallicity. Central LSB and HSB galaxies inhabit similarly low-density large-scale ($>$200 kpc) environments, but LSB galaxies are more isolated on small scales ($\sim$100 kpc). Even after matching in stellar mass and environment, LSB galaxies show systematically lower $Σ_\ast$, $Σ_{SFR}$, and metallicities, often hosting diffuse, weakly star-forming bulges embedded in extended disks. These results indicate that LSB structure and star formation are not primarily governed by large-scale environment or halo mass. While secondary halo properties such as spin, concentration, or gas accretion history are often invoked, their environmental dependence appears weak. Instead, LSB-HSB differences for centrals likely reflect divergent assembly or interaction histories and internal processes -- such as angular momentum-driven disk evolution or inefficient gas conversion -- largely decoupled from large-scale environment. Nonetheless, environment still influences the observed star formation and chemical differences between central and satellite LSB galaxies.

astro-ph.GA↗
arXiv2026

Ecohydrological Controls on Moist Convection and Long-Term Rainfall Feedback

To elucidate how land surface state and soil moisture dynamics regulate moist convection, and how convective rainfall subsequently reshapes surface and root-zone hydrology, we develop a stochastic dynamical model that couples soil moisture, vegetation hydraulics, atmospheric boundary layer evolution, and convective available potential energy (CAPE). We show that CAPE depends not only on the free-tropospheric environment but also on soil moisture, through its control of surface fluxes, boundary-layer growth, and the timing of the intersection between the atmospheric boundary layer and the lifting condensation level (LCL). Soil texture and plant properties strongly modulate convective potential during dry-down. Loamy sand favors convection at relatively high soil moisture and maintains the largest CAPE at the time of LCL-ABL crossing across drying conditions. In contrast, sandy soils exhibit high CAPE when wet but lose convective potential rapidly as they dry. As matric potential becomes more negative, convection is increasingly suppressed in finer, loamy clay textures. Plant functional type further shapes dry-down dynamics: water-use-maximizing strategies can enhance dry persistence via stomatal closure during drying, whereas more conservative strategies can sustain convection for longer periods. On longer timescales, stochastic rainfall forcing with CAPE-dependent precipitation intensity produces persistent wet and dry soil moisture regimes, with switching times that depend on soil hydraulic properties, plant physiological traits, and atmospheric conditions.

physics.ao-ph↗
arXiv2026

Incomplete Multi-View Multi-Label Classification via Shared Codebook and Fused-Teacher Self-Distillation

Although multi-view multi-label learning has been extensively studied, research on the dual-missing scenario, where both views and labels are incomplete, remains largely unexplored. Existing methods mainly rely on contrastive learning or information bottleneck theory to learn consistent representations under missing-view conditions, but loss-based alignment without explicit structural constraints limits the ability to capture stable and discriminative shared semantics. To address this issue, we introduce a more structured mechanism for consistent representation learning: we learn discrete consistent representations through a multi-view shared codebook and cross-view reconstruction, which naturally align different views within the limited shared codebook embeddings and reduce feature redundancy. At the decision level, we design a weight estimation method that evaluates the ability of each view to preserve label correlation structures, assigning weights accordingly to enhance the quality of the fused prediction. In addition, we introduce a fused-teacher self-distillation framework, where the fused prediction guides the training of view-specific classifiers and feeds the global knowledge back into the single-view branches, thereby enhancing the generalization ability of the model under missing-label conditions. The effectiveness of our proposed method is thoroughly demonstrated through extensive comparative experiments with advanced methods on five benchmark datasets. Code is available at https://github.com/xuy11/SCSD.

cs.CV↗
arXiv2026

Spin-based magnetic detection of optically trapped single cell in microfluidic channel

Combining optical tweezers with fluorescence microscopy is a powerful tool for single-cell analysis, playing a pivotal role in disease diagnosis, cell sorting, and the investigation of cellular dynamics. However, fluorescence detection faces challenges such as blinking, photobleaching and autofluorescence in biotissues. To address these limitations, we developed a magnetic detection strategy by integrating quantum magnetometry using nitrogen-vacancy centers into optical tweezers, demonstrating precise trapping and manipulation of individual cells in microfluidic environment. We detected a magnetic signal of 89 μT from a single cell labeled with magnetic nanoparticles, compared to a noise floor of 3.9 μT observed in unlabeled cells. This platform provides a promising approach for high-precision single-cell analysis and holds significant potential for probing cellular activities within biological microenvironments.

physics.optics↗
arXiv2026

Careful Queries, Credible Results: Teaching RAG Models Advanced Web Search Tools with Reinforcement Learning

Retrieval-Augmented Generation (RAG) enhances large language models (LLMs) by integrating up-to-date external knowledge, yet real-world web environments present unique challenges. These limitations manifest as two key challenges: pervasive misinformation in the web environment, which introduces unreliable or misleading content that can degrade retrieval accuracy, and the underutilization of web tools, which, if effectively employed, could enhance query precision and help mitigate this noise, ultimately improving the retrieval results in RAG systems. To address these issues, we propose WebFilter, a novel RAG framework that generates source-restricted queries and filters out unreliable content. This approach combines a retrieval filtering mechanism with a behavior- and outcome-driven reward strategy, optimizing both query formulation and retrieval outcomes. Extensive experiments demonstrate that WebFilter improves answer quality and retrieval precision, outperforming existing RAG methods on both in-domain and out-of-domain benchmarks.

cs.IR↗
arXiv2026

SDSS-IV MaNGA: Distinct Structural Growth and Star Formation in Low and High Surface Brightness Disks

We analyze a clean sample of 1,118 late-type, face-on galaxies without AGN contamination from the MaNGA survey. Their photometric structures are quantified via two-component (bulge+disk) decompositions on deep $g$-band images from the DESI Legacy Survey. Using a disk central surface brightness of $μ_{\rm 0,d,cor}$(g) = 22 $\pm$ 0.3 mag arcsec$^{-2}$ (corrected for inclination and cosmic dimming) as the classification threshold, we identify 159 low surface brightness (LSB) galaxies, 388 LSB candidates, and 571 high surface brightness (HSB) galaxies. LSB galaxies are predominantly low-mass ($M_\ast < 3 \times 10^{10}$ M$_\odot$), exhibiting 29\% larger effective radii, 15\% lower star formation rates (SFRs), and 12\% reduced gas-phase metallicities than HSB counterparts at comparable masses. These differences cause systematic offsets from standard scaling relations. Despite comparable gas content, LSB galaxies host older stellar populations, longer gas depletion times, and less efficient star formation. Spatially resolved analyses further reveal that LSB galaxies display centrally suppressed $Σ_{\rm SFR}$, flatter SFR gradients, and rising specific SFR profiles toward their outskirts. Together with steeper negative metallicity gradients, these trends suggest ongoing gas accretion fueling outer-disk star formation. Consistently, the outer regions of LSB galaxies exhibit stronger H$δ_A$ absorption and lower D$_n$4000 indices, indicating fading A-star populations. Moreover, LSB galaxies show lower $Σ_{\ast}$ across all $R/R_e$ and more centrally depleted stellar mass profiles on an absolute radial scale, compared with HSB and large-size star-forming galaxies. Collectively, LSB galaxies represent a distinct population with slow evolution, inefficient star formation, and continued susceptibility to late-time gas accretion and peripheral star formation.

astro-ph.GA↗
arXiv2026

When to Lock Attention: Training-Free KV Control in Video Diffusion

Maintaining background consistency while enhancing foreground quality remains a core challenge in video editing. Injecting full-image information often leads to background artifacts, whereas rigid background locking severely constrains the model's capacity for foreground generation. To address this issue, we propose KV-Lock, a training-free framework tailored for DiT-based video diffusion models. Our core insight is that the hallucination metric (variance of denoising prediction) directly quantifies generation diversity, which is inherently linked to the classifier-free guidance (CFG) scale. Building upon this, KV-Lock leverages diffusion hallucination detection to dynamically schedule two key components: the fusion ratio between cached background key-values (KVs) and newly generated KVs, and the CFG scale. When hallucination risk is detected, KV-Lock strengthens background KV locking and simultaneously amplifies conditional guidance for foreground generation, thereby mitigating artifacts and improving generation fidelity. As a training-free, plug-and-play module, KV-Lock can be easily integrated into any pre-trained DiT-based models. Extensive experiments validate that our method outperforms existing approaches in improved foreground quality with high background fidelity across various video editing tasks.

cs.CV↗
arXiv2026

Mock Observations for the CSST Mission: Integral Field Spectrograph--GEHONG: A Package for Generating Ideal Datacubes

We developed a Python package GEHONG to mock the three-dimensional spectral data cube under the observation of an ideal telescope for the Integral Field Spectrograph of the Chinese Space Station Telescope (CSST-IFS). This package can generate one-dimensional spectra corresponding to local physical properties at specific positions according to a series of two-dimensional distributions of physical parameters of target sources. In this way, it can produce a spatially resolved spectral cube of the target source. Two-dimensional distributions of physical parameters, including surface brightness, stellar population, and line-of-sight velocity, can be modeled using the parametric model or based on real observational data and numerical simulation data. For the generation of one-dimensional spectra, we have considered four types of spectra, including the stellar continuum spectra, ionized gas emission lines, AGN spectra, and stellar spectra. That makes GEHONG able to mock various types of targets, including galaxies, AGNs, star clusters, and HII regions.

astro-ph.IM↗
arXiv2026

A 16 nm 1.60TOPS/W High Utilization DNN Accelerator with 3D Spatial Data Reuse and Efficient Shared Memory Access

Achieving high compute utilization across a wide range of AI workloads is crucial for the efficiency of versatile DNN accelerators. This paper presents the Voltra chip and its utilization-optimised DNN accelerator architecture, which leverages 3-Dimensional (3D) spatial data reuse along with efficient and flexible shared memory access. The 3D spatial dataflow enables balanced spatial data reuse across three dimensions, improving spatial utilization by up to 2.0x compared to a conventional 2D design. Inside the shared memory access architecture, Voltra incorporates flexible data streamers that enable mixed-grained hardware data pre-fetching and dynamic memory allocation, further improving the temporal utilization by 2.12-2.94x and achieving 1.15-2.36x total latency speedup compared with the non-prefetching and separated memory architecture, respectively. Fabricated in 16nm technology, our chip achieves 1.60 TOPS/W peak system energy efficiency and 1.25 TOPS/mm2 system area efficiency, which is competitive with state-of-the-art solutions while achieving high utilization across diverse workloads.

cs.AR↗
arXiv2026

PS$^2$: Parameterized Control for Fine-Grained Student Proficiency Simulation

Understanding how students with different proficiency levels respond to educational materials is a critical issue within the field of AI for Education. However, acquiring sufficient real student response data for a robust evaluation is often hindered by cost, ethics, and security constraints. Consequently, LLM-based student proficiency simulation, especially prompt-based methods, has emerged as a practical alternative under data-scarce conditions. Despite their promise, current methods still exhibit limited controllability with coarse-grained proficiency representations, high sensitivity to prompt design, and the lack of calibration with academic performance. Therefore, we propose Parameterized Student Proficiency Simulation (PS$^2$), an unsupervised and parameterized model-level framework that simulates students with different proficiencies by interpolating between a strong upper-bound LLM and a weaker, cognitive error-informed lower-bound student LLM via a hybrid ratio. Specifically, the lower-bound model is constructed by fine-tuning the weaker LM to exhibit cognitive errors when responding to educational materials. To ensure alignment with target proficiency levels, PS$^2$ further calibrates the interpolation ratio with academic performance. Experiments on two public datasets demonstrate that PS$^2$ achieves finer-grained and consistent proficiency simulation compared to existing baselines, leading to superior performance in student behavior similarity and item difficulty prediction.

cs.CY↗