Search arXiv⌕ Search

arXiv subjects

Kai Wang

Publications and source records attributed to Kai Wang.

At least 199 records · Page 11Linked to original sources
arXiv2026

A census of quiescent galaxies across $0.5 < z < 8$ with JWST/MIRI: Mass-dependent number density evolution of quiescent galaxies in the early Universe

Recent JWST observations have revealed a large population of quiescent galaxies (QGs) at high redshift ($z \sim 4-8$), challenging current models of early galaxy formation and quenching. Accurate number density estimates are crucial but remain uncertain. We present a systematic study of QGs at $0.5 < z < 8$ using a mass-complete sample from the JWST/PRIMER survey with deep NIRCam and MIRI imaging. We demonstrate that MIRI photometry is essential for refining the QG sample: it helps to mitigate contamination from dusty star-forming galaxies in the high-mass regime at $z \sim 3-5$ and aids in recovering lower-mass QG candidates at $z > 5$ that are often missed without including MIRI data. We find that the evolution of the QG number density is strongly mass-dependent. The density of massive QGs ($\log (M_{\star}/M_{\odot}) > 10.6$) declines rapidly, falling from $n \approx 1.32\times10^{-5}~~\mathrm{Mpc^{-3}}$ at $z \sim 3-4$ to $n < 1 \times10^{-6}~~\mathrm{Mpc^{-3}}$ at $z \sim 6$, and becomes negligible at $z > 6$. In contrast, low-mass QGs ($9.5 < \log (M_{\star}/M_{\odot}) < 10.6$) exhibit a remarkably constant number density of $n \sim 3\times10^{-6}~\mathrm{Mpc^{-3}}$ across the redshift range $z = 4-8$. This plateau suggests that these high-redshift, low-mass QGs may be galaxies undergoing temporary quenching episodes, likely subject to rejuvenation upon future gas accretion. Comparisons with leading galaxy formation models reveal significant tensions: most models underestimate the abundance of massive QGs at $z > 4$ and fail to reproduce the flat density evolution observed for the low-mass population.

astro-ph.GA↗
arXiv2026

From Oracle to Noisy Context: Mitigating Contextual Exposure Bias in Speech-LLMs

Contextual automatic speech recognition (ASR) with Speech-LLMs is typically trained with oracle conversation history, but relies on error-prone history at inference, causing a train-test mismatch in the context channel that we term contextual exposure bias. We propose a unified training framework to improve robustness under realistic histories: (i) Teacher Error Knowledge by using Whisper large-v3 hypotheses as training-time history, (ii) Context Dropout to regularize over-reliance on history, and (iii) Direct Preference Optimization (DPO) on curated failure cases. Experiments on TED-LIUM 3 (in-domain) and zero-shot LibriSpeech (out-of-domain) show consistent gains under predicted-history decoding. With a two-utterance history as context, SFT with Whisper hypotheses reduce WER from 5.59% (oracle-history training) to 5.47%, and DPO further improves to 5.17%. Under irrelevant-context attacks, DPO yields the smallest degradation (5.17% -> 5.63%), indicating improved robustness to misleading context. Our code and models are published on https://github.com/XYGuo1996/Contextual_Speech_LLMs.

cs.CL↗
arXiv2026

DI3CL: Contrastive Learning With Dynamic Instances and Contour Consistency for SAR Land-Cover Classification Foundation Model

Although significant advances have been achieved in SAR land-cover classification, recent methods remain predominantly focused on supervised learning, which relies heavily on extensive labeled datasets. This dependency not only limits scalability and generalization but also restricts adaptability to diverse application scenarios. In this paper, a general-purpose foundation model for SAR land-cover classification is developed, serving as a robust cornerstone to accelerate the development and deployment of various downstream models. Specifically, a Dynamic Instance and Contour Consistency Contrastive Learning (DI3CL) pre-training framework is presented, which incorporates a Dynamic Instance (DI) module and a Contour Consistency (CC) module. DI module enhances global contextual awareness by enforcing local consistency across different views of the same region. CC module leverages shallow feature maps to guide the model to focus on the geometric contours of SAR land-cover objects, thereby improving structural discrimination. Additionally, to enhance robustness and generalization during pre-training, a large-scale and diverse dataset named SARSense, comprising 460,532 SAR images, is constructed to enable the model to capture comprehensive and representative features. To evaluate the generalization capability of our foundation model, we conducted extensive experiments across a variety of SAR land-cover classification tasks, including SAR land-cover mapping, water body detection, and road extraction. The results consistently demonstrate that the proposed DI3CL outperforms existing methods. Our code and pre-trained weights are publicly available at: https://github.com/SARpre-train/DI3CL.

cs.CV↗
arXiv2026

Detecting AI-Generated Content in Academic Peer Reviews

The growing availability of large language models (LLMs) has raised questions about their role in academic peer review. This study examines the temporal emergence of AI-generated content in peer reviews by applying a detection model trained on historical reviews to later review cycles at International Conference on Learning Representations (ICLR) and Nature Communications (NC). We observe minimal detection of AI-generated content before 2022, followed by a substantial increase through 2025, with approximately 20% of ICLR reviews and 12% of Nature Communications reviews classified as AI-generated in 2025. The most pronounced growth of AI-generated reviews in NC occurs between the third and fourth quarter of 2024. Together, these findings provide suggestive evidence of a rapidly increasing presence of AI-assisted content in peer review and highlight the need for further study of its implications for scholarly evaluation.

cs.CL↗
arXiv2026

Approximate Butterfly Counting in Sublinear Time

Bipartite graphs serve as a natural model for representing relationships between two different types of entities. When analyzing bipartite graphs, butterfly counting is a fundamental research problem that aims to count the number of butterflies (i.e., 2x2 bicliques) in a given bipartite graph. While this problem has been extensively studied in the literature, existing algorithms usually necessitate access to a large portion of the entire graph, presenting challenges in real scenarios where graphs are extremely large and I/O costs are expensive. In this paper, we study the butterfly counting problem under the query model, where the following query operations are permitted: degree query, neighbor query, and vertex-pair query. We propose TLS, a practical two-level sampling algorithm that can estimate the butterfly count accurately while accessing only a limited graph structure, achieving significantly lower query costs under the standard query model. TLS also incorporates several key techniques to control the variance, including "small-degree-first sampling" and "wedge sampling via small subsets". To ensure theoretical guarantees, we further introduce two novel techniques: "heavy-light partition" and "guess-and-prove", integrated into TLS. With these techniques, we prove that the algorithm can achieve a (1+eps) accuracy for any given approximation parameter 0 < eps < 1 on general bipartite graphs with a promised time and query complexity. In particular, the promised time is sublinear when the input graph is dense enough. Extensive experiments on 15 datasets demonstrate that TLS delivers robust estimates with up to three orders of magnitude lower query costs and runtime compared to existing solutions.

cs.DB↗
arXiv2026

Toward Real-Time Surgical Scene Segmentation via a Spike-Driven Video Transformer with Spike-Informed Pretraining

Modern surgical systems increasingly rely on intelligent scene understanding to improve intra-operative safety and situational awareness, with surgical scene segmentation playing a fundamental role in fine-grained surgical perception. Although recent ANN models, especially large foundation models, have achieved impressive accuracy, their high computational and energy demands often hinder deployment in resource-constrained operative environments. To address this challenge, we explore SNN as a highly efficient paradigm. However, its performance in surgical scene segmentation remains constrained by sparse spike representations and limited annotated surgical data. We therefore propose SpikeSurgSeg, the first spike-driven video Transformer for surgical scene segmentation. It preserves the real-time and energy-efficient advantages of SNN, while achieving competitive performance against most ANN models in data-scarce surgical scenarios. Specifically, we introudce a spike-informed pretraining strategy based on MAE, where mask generation is guided by spike firing activity to better align with sparse spike representations, together with a layer-wise tube masking scheme that reduces information leakage and encourages contextual reasoning. To further strengthen semantic representation, we introduce multi-spectral knowledge distillation, which aligns teacher ANN and student SNN features in the frequency domain, where the mismatch between continuous activation patterns and spike-driven temporal representations can be effectively mitigated. Built on the pretrained SNN encoder, we further design a lightweight spike-driven segmentation head. Extensive experiments on EndoVis18 and our in-house SurgBleed dataset show that SpikeSurgSeg achieves mIoU comparable to SOTA ANN models while reducing inference latency by at least 8x. Notably, it delivers over 20x speedup relative to most foundation models.

cs.CV↗
arXiv2026

Chain-of-Trajectories: Unlocking the Intrinsic Generative Optimality of Diffusion Models via Graph-Theoretic Planning

Diffusion models operate in a reflexive System 1 mode, constrained by a fixed, content-agnostic sampling schedule. This rigidity arises from the curse of state dimensionality, where the combinatorial explosion of possible states in the high-dimensional noise manifold renders explicit trajectory planning intractable and leads to systematic computational misallocation. To address this, we introduce Chain-of-Trajectories (CoTj), a train-free framework enabling System 2 deliberative planning. Central to CoTj is Diffusion DNA, a low-dimensional signature that quantifies per-stage denoising difficulty and serves as a proxy for the high-dimensional state space, allowing us to reformulate sampling as graph planning on a directed acyclic graph. Through a Predict-Plan-Execute paradigm, CoTj dynamically allocates computational effort to the most challenging generative phases. Experiments across multiple generative models demonstrate that CoTj discovers context-aware trajectories, improving output quality and stability while reducing redundant computation. This work establishes a new foundation for resource-aware, planning-based diffusion modeling. The code is available at https://github.com/UnicomAI/CoTj.

cs.LG↗
arXiv2026

Advancing Multimodal Agent Reasoning with Long-Term Neuro-Symbolic Memory

Recent advances in large language models have driven the emergence of intelligent agents operating in open-world, multimodal environments. To support long-term reasoning, such agents are typically equipped with external memory systems. However, most existing multimodal agent memories rely primarily on neural representations and vector-based retrieval, which are well-suited for inductive, intuitive reasoning but fundamentally limited in supporting analytical, deductive reasoning critical for real-world decision making. To address this limitation, we propose NS-Mem, a long-term neuro-symbolic memory framework designed to advance multimodal agent reasoning by integrating neural memory with explicit symbolic structures and rules. Specifically, NS-Mem is operated around three core components of a memory system: (1) a three-layer memory architecture that consists episodic layer, semantic layer and logic rule layer, (2) a memory construction and maintenance mechanism implemented by SK-Gen that automatically consolidates structured knowledge from accumulated multimodal experiences and incrementally updates both neural representations and symbolic rules, and (3) a hybrid memory retrieval mechanism that combines similarity-based search with deterministic symbolic query functions to support structured reasoning. Experiments on real-world multimodal reasoning benchmarks demonstrate that Neural-Symbolic Memory achieves an average 4.35% improvement in overall reasoning accuracy over pure neural memory systems, with gains of up to 12.5% on constrained reasoning queries, validating the effectiveness of NS-Mem.

cs.AI↗
arXiv2026

TrinityGuard: A Unified Framework for Safeguarding Multi-Agent Systems

With the rapid development of LLM-based multi-agent systems (MAS), their significant safety and security concerns have emerged, which introduce novel risks going beyond single agents or LLMs. Despite attempts to address these issues, the existing literature lacks a cohesive safeguarding system specialized for MAS risks. In this work, we introduce TrinityGuard, a comprehensive safety evaluation and monitoring framework for LLM-based MAS, grounded in the OWASP standards. Specifically, TrinityGuard encompasses a three-tier fine-grained risk taxonomy that identifies 20 risk types, covering single-agent vulnerabilities, inter-agent communication threats, and system-level emergent hazards. Designed for scalability across various MAS structures and platforms, TrinityGuard is organized in a trinity manner, involving an MAS abstraction layer that can be adapted to any MAS structures, an evaluation layer containing risk-specific test modules, alongside runtime monitor agents coordinated by a unified LLM Judge Factory. During Evaluation, TrinityGuard executes curated attack probes to generate detailed vulnerability reports for each risk type, where monitor agents analyze structured execution traces and issue real-time alerts, enabling both pre-development evaluation and runtime monitoring. We further formalize these safety metrics and present detailed case studies across various representative MAS examples, showcasing the versatility and reliability of TrinityGuard. Overall, TrinityGuard acts as a comprehensive framework for evaluating and monitoring various risks in MAS, paving the way for further research into their safety and security.

cs.CR↗
arXiv2026

Memory-Driven Role-Playing: Evaluation and Enhancement of Persona Knowledge Utilization in LLMs

A core challenge for faithful LLM role-playing is sustaining consistent characterization throughout long, open-ended dialogues, as models frequently fail to recall and accurately apply their designated persona knowledge without explicit cues. To tackle this, we propose the Memory-Driven Role-Playing paradigm. Inspired by Stanislavski's "emotional memory" acting theory, this paradigm frames persona knowledge as the LLM's internal memory store, requiring retrieval and application based solely on dialogue context, thereby providing a rigorous test of depth and autonomous use of knowledge. Centered on this paradigm, we contribute: (1) MREval, a fine-grained evaluation framework assessing four memory-driven abilities - Anchoring, Recalling, Bounding, and Enacting; (2) MRPrompt, a prompting architecture that guides structured memory retrieval and response generation; and (3) MRBench, a bilingual (Chinese/English) benchmark for fine-grained diagnosis. The novel paradigm provides a comprehensive diagnostic for four-staged role-playing abilities across 12 LLMs. Crucially, experiments show that MRPrompt allows small models (e.g., Qwen3-8B) to match the performance of much larger closed-source LLMs (e.g., Qwen3-Max and GLM-4.7), and confirms that upstream memory gains directly enhance downstream response quality, validating the staged theoretical foundation.

cs.CL↗
arXiv2026

NumColor: Precise Numeric Color Control in Text-to-Image Generation

Text-to-image diffusion models excel at generating images from natural language descriptions, yet fail to interpret numerical colors such as hex codes (#FF5733) and RGB values (rgb(255,87,51)). This limitation stems from subword tokenization, which fragments color codes into semantically meaningless tokens that text encoders cannot map to coherent color representations. We present NumColor, that enables precise numerical color control across multiple diffusion architectures. NumColor comprises two components: a Color Token Aggregator that detects color specifications regardless of tokenization, and a ColorBook containing 6,707 learnable embeddings that map colors to embedding space of text encoder in perceptually uniform CIE Lab space. We introduce two auxiliary losses, directional alignment and interpolation consistency, to enforce geometric correspondence between Lab and embedding spaces, enabling smooth color interpolation. To train the ColorBook, we construct NumColor-Data, a synthetic dataset of 500K rendered images with unambiguous color-to-pixel correspondence, eliminating the annotation ambiguity inherent in photographic datasets. Although trained solely on FLUX, NumColor transfers zero-shot to SD3, SD3.5, PixArt-α, and PixArt-Σ without model-specific adaptation. NumColor improves numerical color accuracy by 4-9x across five models, while simultaneously improving color harmony scores by 10-30x on GenColorBench benchmark.

cs.CV↗
arXiv2026

Atomic data benchmarked by Large-scale Multiconfiguration Dirac-Hartree-Fock Calculations for Beryllium

The multiconfiguration Dirac-Hartree-Fock (MCDHF) and relativistic configuration interaction (RCI) methods are used to provide excitation energies, radiative transition data, lifetimes, Lande g-factors, hyperfine interaction constants and isotope shift parameters for the 99 lowest levels of configurations 1s^22snl (n <= 7) + 1s^22p^2 in beryllium. Compared with available experimental excitation energies, the average difference with the standard deviation is 7.08 +/- 1.14cm^-1 (0.011% +/- 0.003%), which demonstrates the excellent theory-observation agreement. The uncertainties of the transition rates are estimated based on two independent methods. The present MCDHF/RCI oscillator strengths and those obtained from the explicitly correlated Gaussian (ECG) method all agree within 2%, except for four transitions affected by strong cancellation effects. For lifetimes, hyperfine splittings and isotope shifts, the present MCDHF/RCI results show good agreement with the few available experimental values, supporting the reliability of our predictions for many states lacking experimental measurements. These comprehensive results can be used in line identification and diagnostics of astrophysical plasmas.

physics.atom-ph↗
arXiv2026

Development of Readout Electronics for a High-Speed Event-Driven Neutron Imaging Detector Based on Timepix4

As the Chinese Spallation Neutron Source enters Phase II, the increase in proton beam power will lead to a further boost in the intensity of pulsed neutron beams. To address the demand for higher event-rate readout electronics for energy-resolved neutron imaging detectors, we have developed a high-performance readout electronics system based on the Timepix4 chip. The prototype electronics system comprises a Timepix4 chip board and a high-performance digital board, which are interconnected through a custom FMC interface. The advantage of this system is its ability to achieve the full bandwidth readout of 160 Gbps for a single Timepix4 chip. The electronics system, based solely on a single ZYNQ-MPSOC chip, is capable of fully meeting the required performance specifications within a compact form factor of 8 cm x 30 cm. Furthermore, the system features a high-capacity external SODIMM memory interface (supporting up to 32 GB), which ensures stable data readout through a single 40 Gbps QSFP+ interface. As of the present moment, notable progress has been achieved, including the successful establishment of 16 data channels between Timepix4 and FPGA that operate error-free and stably at a speed of 5.12 Gbps, which is half of the maximum theoretical speed of 10.24 Gbps. The threshold standard deviation across all pixels is less than 50 e- after equalization. And the clear structural results obtained from X-ray experiments indicate that the functionality is essentially complete, allowing further testing.

physics.ins-det↗
arXiv2026

MeanCache: From Instantaneous to Average Velocity for Accelerating Flow Matching Inference

We present MeanCache, a training-free caching framework for efficient Flow Matching inference. Existing caching methods reduce redundant computation but typically rely on instantaneous velocity information (e.g., feature caching), which often leads to severe trajectory deviations and error accumulation under high acceleration ratios. MeanCache introduces an average-velocity perspective: by leveraging cached Jacobian--vector products (JVP) to construct interval average velocities from instantaneous velocities, it effectively mitigates local error accumulation. To further improve cache timing and JVP reuse stability, we develop a trajectory-stability scheduling strategy as a practical tool, employing a Peak-Suppressed Shortest Path under budget constraints to determine the schedule. Experiments on FLUX.1, Qwen-Image, and HunyuanVideo demonstrate that MeanCache achieves 4.12X and 4.56X and 3.59X acceleration, respectively, while consistently outperforming state-of-the-art caching baselines in generation quality. We believe this simple yet effective approach provides a new perspective for Flow Matching inference and will inspire further exploration of stability-driven acceleration in commercial-scale generative models.

cs.LG↗
arXiv2026

Beyond Geometry: Artistic Disparity Synthesis for Immersive 2D-to-3D

Current 2D-to-3D conversion methods achieve geometric accuracy but are artistically deficient, failing to replicate the immersive and emotionally resonant experience of professional 3D cinema. This is because geometric reconstruction paradigms mistake deliberate artistic intent, such as strategic zero-plane shifts for pop-out effects and local depth sculpting, for data noise or ambiguity. This paper argues for a new paradigm: Artistic Disparity Synthesis, shifting the goal from physically accurate disparity estimation to artistically coherent disparity synthesis. We propose Art3D, a preliminary framework exploring this paradigm. Art3D uses a dual-path architecture to decouple global depth parameters (macro-intent) from local artistic effects (visual brushstrokes) and learns from professional 3D film data via indirect supervision. We also introduce a preliminary evaluation method to quantify cinematic alignment. Experiments show our approach demonstrates potential in replicating key local out-of-screen effects and aligning with the global depth styles of cinematic 3D content, laying the groundwork for a new class of artistically-driven conversion tools.

cs.CV↗
arXiv2026

Orbital-Selective Spin-Orbit Mott Insulator in Fractional Valence Iridate La$_3$Ir$_3$O$_{11}$

The combination of strong spin-orbit coupling and Coulomb interactions makes the $5d$ iridates a unique platform for realizing novel correlated electronic states. Here, utilizing infrared spectroscopy, we demonstrate that a robust Mott insulating state persists in the $1/3$-hole self-doped system La$_3$Ir$_3$O$_{11}$, evidenced by the collapse of the Drude response and the emergence of sharp excitations across the Mott gap. Our theoretical calculations reveal that the insulating behavior arises from the cooperative interplay of structural distortions, spin-orbit coupling, and Coulomb interactions. Specifically, octahedral distortion and Ir-Ir dimerization split the $t_{2g}$ orbitals, driving the $J_{\mathrm{eff}} = 1/2$ bands toward half-filling while keeping the $J_{\mathrm{eff}} = 3/2$ bands away from it. Consequently, electron correlations induce an orbital-selective Mott transition in the $J_{\mathrm{eff}} = 1/2$ bands, whereas a band-insulating gap develops in the $J_{\mathrm{eff}} = 3/2$ bands, thereby stabilizing the unconventional insulating state in La$_3$Ir$_3$O$_{11}$. These findings provide new insights into the design and understanding of the insulating ground state of spin-orbit-coupled iridates.

cond-mat.str-el↗
arXiv2026

Towards Principled Dataset Distillation: A Spectral Distribution Perspective

Dataset distillation (DD) aims to compress large-scale datasets into compact synthetic counterparts for efficient model training. However, existing DD methods exhibit substantial performance degradation on long-tailed datasets. We identify two fundamental challenges: heuristic design choices for distribution discrepancy measure and uniform treatment of imbalanced classes. To address these limitations, we propose Class-Aware Spectral Distribution Matching (CSDM), which reformulates distribution alignment via the spectrum of a well-behaved kernel function. This technique maps the original samples into frequency space, resulting in the Spectral Distribution Distance (SDD). To mitigate class imbalance, we exploit the unified form of SDD to perform amplitude-phase decomposition, which adaptively prioritizes the realism in tail classes. On CIFAR-10-LT, with 10 images per class, CSDM achieves a 14.0% improvement over state-of-the-art DD methods, with only a 5.7% performance drop when the number of images in tail classes decreases from 500 to 25, demonstrating strong stability on long-tailed data.

cs.CV↗
arXiv2026

Mixture of Sequence: Theme-Aware Mixture-of-Experts for Long-Sequence Recommendation

Sequential recommendation has rapidly advanced in click-through rate prediction due to its ability to model dynamic user interests. A key challenge, however, lies in modeling long sequences: users often exhibit significant interest shifts, introducing substantial irrelevant or misleading information. Our empirical analysis corroborates this challenge and uncovers a recurring behavioral pattern in long sequences (\textit{session hopping}): user interests remain stable within short temporal spans (\textit{sessions}) but shift drastically across sessions and may reappear after multiple sessions. To address this challenge, we propose the Mixture of Sequence (MoS) framework, a model-agnostic MoE approach that achieves accurate predictions by extracting theme-specific and multi-scale subsequences from noisy raw user sequences. First, MoS employs a theme-aware routing mechanism to adaptively learn the latent themes of user sequences and organizes these sequences into multiple coherent subsequences. Each subsequence contains only sessions aligned with a specific theme, thereby effectively filtering out irrelevant or even misleading information introduced by user interest shifts in session hopping. In addition, to alleviate potential information loss, we introduce a multi-scale fusion mechanism, which leverages three types of experts to capture global sequence characteristics, short-term user behaviors, and theme-specific semantic patterns. Together, these two mechanisms endow MoS with the ability to deliver accurate recommendations from multi-faceted and multi-scale perspectives. Experimental results demonstrate that MoS consistently achieves the SOTA performance while introducing fewer FLOPs compared with other MoE counterparts, providing strong evidence of its excellent balance between utility and efficiency. The code is available at https://github.com/xiaolin-cs/MoS.

cs.IR↗