Searcharxiv⌕ Search

arXiv subjects

Junjie Wang

Publications and source records attributed to Junjie Wang.

At least 55 records · Page 3Linked to original sources
arXiv2026

Skim and Skip: Hierarchical Adaptive Inference for Efficient Multimodal Retrieval

Universal multimodal retrieval (UMR) increasingly adopts multimodal large language models (MLLMs) as unified embedding backbones, but their strong retrieval performance comes at substantial inference cost. Existing methods typically rely on uniformly dense inference, where all input tokens are processed through the entire model and matched using the final-layer [EOS] representation. However, this paradigm overlooks two key forms of heterogeneity in multimodal retrieval: token contributions to the final retrieval embedding are highly uneven, and different queries require markedly different amounts of inference depth. To address this, we propose Skim and Skip (SAS), a hierarchical adaptive inference framework for efficient multimodal retrieval. SAS first performs token-level evidence selection to preserve only the input information most relevant to the final retrieval embedding, and then performs depth-adaptive inference to determine whether the current representation is already sufficient for reliable matching. Experiments on 12 MMEB retrieval tasks show that SAS retains about 99% of the dense baseline's average retrieval performance while achieving up to 1.64 times end-to-end speedup and up to 66.3% FLOPs reduction.

cs.IR↗
arXiv2026

Agile Fall Recovery for Quadrotors with Bidirectional Thrust via Reinforcement Learning

Autonomous fall recovery is a critical capability for quadrotors operating in real-world environments, where collisions or failures may leave the vehicle resting on the ground in an arbitrary attitude. This problem is challenging because recovery must be achieved under limited onboard sensing, in constrained free space, with ground contact, and in the presence of unknown disturbances. In this letter, we present an RL-based framework for autonomous fall recovery of a quadrotor from arbitrary ground attitudes to stable hover using only lightweight onboard sensors. To address severe partial observability and intermittent sensor invalidity, we train a recurrent policy within an asymmetric actor--critic architecture, leveraging an Incremental Nonlinear Dynamic Inversion (INDI) controller to track the policy output. Combined with high-fidelity simulations of motor response and optical flow, the overall training framework significantly reduces the sim-to-real gap. Simulation ablation studies validate the importance of the main design choices, while real-world experiments demonstrate zero-shot transfer and robust recovery under different initial attitudes, wind disturbances, and additional payloads. These results demonstrate that agile quadrotor fall recovery can be achieved without explicit state estimation using only limited and unreliable onboard sensing.

cs.RO↗
arXiv2026

EurekAgent: Agent Environment Engineering is All You Need For Autonomous Scientific Discovery

LLM-based agents have shown increasing potential in automating scientific discovery. Given an optimizable metric and an execution environment, they can propose, validate, and iterate scientific solutions, and have produced results that outperform human-designed approaches. As model capabilities continue to improve, we argue that the bottleneck for autonomous scientific discovery is shifting from prescribing agent workflows to designing agent environments: the resources, constraints, and interfaces that shape agent behavior. We frame this as environment engineering: building environments that amplify productive behaviors, such as open-ended exploration, systematic artifact management, and inter-agent collaboration, while suppressing harmful behaviors, such as reward hacking and high-friction human oversight. We present EurekAgent, an environment-engineered agent system for metric-driven autonomous scientific discovery. EurekAgent engineers the environment along four dimensions: permissions engineering for bounded agent execution and isolated evaluation; artifact engineering for filesystem and Git-based collaboration; budget engineering for budget-aware exploration; and human-in-the-loop engineering for easy human supervision and intervention. EurekAgent sets new state-of-the-art results on multiple mathematics, kernel engineering, and machine learning tasks, including new state-of-the-art 26-circle packing results discovered with less than $11 in total API cost. We open-source our code and results, and call for environment engineering as a core research direction for developing reliable autonomous research agents.

cs.AI↗
arXiv2026

The maximum number of edges in $C_{2k+1}^{-}$-free unbalanced signed graphs with bounded clique number

Recently, the Turán problem for graphs with bounded clique number has attracted considerable attention. Since a graph can be regarded as a signed graph without negative edges, it is natural to extend the study of such Turán-type problems to signed graphs. With this motivation, we investigate the Turán-type problem for unbalanced $C_{2k+1}^{-}$-free signed graphs with bounded clique number. In fact, we establish a general result via the classical stability theorem. Specifically, for sufficiently large $n$ and a color-critical graph $F$, we determine the maximum number of edges among all $n$-vertex $C_{2k+1}^{-}$-free unbalanced signed graphs whose underlying graphs are $F$-free.

math.CO↗
arXiv2026

SlimSearcher: Training Efficiency-Aware Web Agents via Adaptive Reward Gating

Deep research agents have demonstrated remarkable capabilities in complex information-seeking tasks, yet this power comes at a steep computational cost. Driven by accuracy-focused training paradigms, current models adopt brute-force strategies characterized by blind tool dependency and performative reasoning-generating long, redundant trajectories that are far from necessary for resolving these tasks, leading to wasteful tool calls and excessive token consumption. To overcome this efficiency trap, we propose SlimSearcher, a principled framework that pushes the Pareto frontier between accuracy and computational cost across both Supervised Fine-Tuning (SFT) and Reinforcement Learning (RL). In the SFT stage, SlimSearcher employs Pareto-efficient filtration to distill trajectories that are both successful and economical, guiding the model toward inherently efficiency-aware search behaviors. During RL, we introduce Adaptive Reward Gating, a dynamic reward-shaping mechanism that evaluates relative tool and token efficiency within a sampled cohort. By cascading these adaptive efficiency metrics with a strict correctness gate, our approach effectively avoids the brevity bias associated with absolute penalties and mitigates reward hacking. Extensive experiments on long-horizon benchmarks, including GAIA, BrowseComp, and XBenchDeepSearch, demonstrate that SlimSearcher reduces average tool-call rounds by 17%-58% while maintaining or improving accuracy.

cs.LG↗
arXiv2026

WebRISE: Requirement-Induced State Evaluation for MLLM-Generated Web Artifacts

Existing benchmarks for MLLM-generated web artifacts assess interaction through local evidence and miss the requirement-induced states and transitions that determine whether a page works. We introduce WebRISE, which compiles task requirements into Interaction Contract Graphs (ICGs) of observable states, user-intent transitions, and DOM/visual assertions for implementation-agnostic browser execution. WebRISE spans 442 tasks across five input modalities (Text, Markdown, Sketch, Image, Video), with 5,495 transitions and 5,271 requirement checks that separate user-stated functions from implicit product-level constraints. Across 14 MLLMs, even the strongest model reaches only 65.6% transition validity and 66.3% requirement coverage, and visual quality is no proxy for behavior (Qwen3.6-35B-A3B on Markdown: V=80.8 yet T=15.5). Video gives the strongest interaction signal (+10.6 pp implicit coverage over Text), while implicit constraints persist; defect injection shows ICG-based scoring detects state errors at 2-16x the rate of checkpoint-style evaluation.

cs.CL↗
arXiv2026

Differentiable Particle-Mesh Ewald with Cartesian Tensor Message Passing for Learning Long-Range Electrostatics and Dipole Response

Machine learning interatomic potentials (MLIPs) can approach quantum accuracy for short-range chemistry, but most architectures remain local and fail to capture the long-range electrostatic and polarization interactions essential for ionic, polar, and interfacial systems. Recent Ewald-based MLIPs show that locally predicted electrostatic variables can recover important long-range physics, including multipolar response. However, many energy-based implementations still compute reciprocal-space terms by direct summation over k vectors, leaving a gap with production molecular dynamics, where particle-mesh Ewald (PME) with O(NlogN) scaling is standard. Here we introduce a fully differentiable PME framework for learned charges and learned atomic dipoles within an E(n)-equivariant Cartesian tensor message passing network. Charges are predicted from scalar local features, while dipoles are predicted from equivariant vector features and enter the same particle-mesh solver as an effective bound charge density. This dipolar density is constructed using analytic real-space gradients of Hockney-Eastwood spline assignment weights, enabling charge-dipole and dipole-dipole long-range forces to be trained end-to-end through FFT-space electrostatics without direct charge or dipole supervision. On a charged-dimer test case, the differentiable PME module reproduces explicit Ewald energies and forces to numerical precision when assignment-kernel deconvolution is enabled. On molten NaCl, the charge and dipole long-range channel gives the lowest force RMSE among the tested models, while all energy RMSE values remain in the sub-meV per atom regime. Timing tests show the expected crossover from explicit Ewald summation to particle-mesh scaling. These results establish differentiable dipole PME as a scalable route toward polarization-aware MLIPs for condensed-phase and interfacial systems.

physics.comp-ph↗
arXiv2026

Decoupled Residual Quantization for Robust Semantic IDs in Recommendation

Semantic IDs represent items as shared discrete token sequences and have become a practical tool for recommendation and retrieval. Yet it remains difficult to tell why a tokenizer fails: poor quality may come from codebook underutilization, unstable decision boundaries, or geometric distortion of the embedding space. This paper develops a quantitative framework for diagnosing these failures through expected codeword overlap and effective codebook capacity. The former measures expected codeword confusion under retrieval-time perturbation, while the latter converts that confusion into an effective number of usable, well-separated codes. The framework links semantic boundary confusion to both code usage imbalance and Euclidean geometric constraints. As a proof of concept, we present Decoupled Residual Quantization (DRQ), which separates continuous geometry reconstruction from discrete distribution matching. Experiments on a large-scale industrial dataset show that Semantic ID quality is multi-objective: symbolic robustness, reconstruction fidelity, and behavior-aware soft matching each stress different aspects of a tokenizer. These downstream observations are based on one proprietary industrial dataset, so they should be read as a case study rather than a universal benchmark claim.

cs.IR↗
arXiv2026

Identifying sensitivity-dominant parameters via active subspaces in reduced-order modeling of fluid dynamics

Reduced-order models (ROMs) are widely employed to describe complex system dynamics when simulations with full-order models (FOMs) are computationally prohibitive. This study presents POD-AS-PRS, a novel model-reduction framework based on the active subspaces (AS) technique, which performs dimensionality reduction in both the state and parameter spaces, enabling efficient and high-fidelity approximations of quantities of interest (QoI). The approach employs proper orthogonal decomposition (POD) to extract low-dimensional coefficients from CFD snapshots, which are inputs to a residual neural network (ResNet) with linear layers to learn their nonlinear mapping to QoI. Reverse-mode automatic differentiation (AD) is utilized to compute gradients with respect to the coefficients, enabling AS analysis to identify influential modes by shifting the analysis to the POD coefficient space, thereby achieving a dual-stage dimensionality reduction driven by QoI sensitivity rather than modal energy. A surrogate model is subsequently constructed using a polynomial response surface (PRS) based on AS-derived active variables, retaining only the highly influential POD coefficients to ensure accurate and efficient QoI reconstruction. The framework is validated on periodic and chaotic bluff-body flows, demonstrating high accuracy with few influential parameters, while AD-based gradients achieve a two-order-of-magnitude speed-up over finite-difference approximations. Sensitivity analysis further reveals that the influential coefficients are not necessarily proportional to modal energy, highlighting the critical flow structures. Consequently, POD-AS-PRS identifies a low-dimensional manifold of sensitivity-dominant parameters that govern the QoI, elucidating the essential flow structures and their coupling with control parameters, thereby enabling efficient and accurate QoI reconstruction.

physics.flu-dyn↗
arXiv2026

MolPIF: A Parameter Interpolation Flow Model for Molecule Generation

Motivation: Structure-based drug design (SBDD) has advanced with deep generative models, but bridging the gap between continuous atomic coordinates and discrete atom types remains a challenge. Current approaches, such as diffusion and flow matching models, often fail to unify these heterogeneous modalities, relying on separate strategies or ill-fitting Euclidean metrics for discrete variables. This lack of a consistent framework limits generative models' ability to capture the geometric and chemical structure of protein-ligand complexes. Results: We present MolPIF, a parameter interpolation flow mechanism designed to unify the generation of continuous and discrete molecular variables. Unlike traditional flow models that operate in sample space, MolPIF interpolates between distributions in the parameter space, theoretically recovering Wasserstein-2 optimal transport for continuous coordinates and establishing Fisher-Rao geodesics for discrete atom types. We further incorporate a geometry-enhanced learning strategy to improve the capture of atomic contexts. Extensive evaluations on the CrossDocked2020 dataset demonstrate that MolPIF outperforms baselines in binding affinity, chemical validity, geometric fidelity and chemical space coverage. Additionally, MolPIF exhibits versatility in lead optimization and offers flexible prior distribution selection (such as Laplace), establishing a robust paradigm for SBDD. Availability: Source code is freely available at https://github.com/BLEACH366/MolPIF. Supplementary information: Supplementary data are available at Bioinformatics.

cs.LG↗
arXiv2026

EvoEGF-Mol: Evolving Exponential Geodesic Flow for Structure-based Drug Design

Structure-Based Drug Design (SBDD) aims to discover bioactive ligands. Conventional approaches construct probability paths separately in Euclidean and probabilistic spaces for continuous atomic coordinates and discrete chemical categories, leading to a mismatch with the underlying statistical manifolds. We address this issue by representing molecules using composite exponential-family distributions, where coordinates and categories are represented within a unified natural parameter space to evolve synchronously along exponential geodesics under the Fisher-Rao metric. To avoid the instantaneous trajectory collapse induced by geodesics directly targeting Dirac distributions, we propose Evolving Exponential Geodesic Flow for SBDD (EvoEGF-Mol), which replaces static Dirac targets with dynamically concentrating distributions and is trained with a progressive-parameter-refinement architecture. Our model approaches a reference-level PoseBusters passing rate (93.4%) on CrossDock, demonstrating remarkable geometric precision and interaction fidelity, while achieving superior performance over baseline methods on real-world MolGenBench tasks for bioactive scaffold recovery. Code is available at https://github.com/BLEACH366/EvoEGF-Mol.

cs.LG↗
arXiv2026

Multi-needle Localization for Pelvic Seed Implant Brachytherapy based on Tip-handle Detection and Matching

Accurate multi-needle localization in intraoperative CT images is crucial for optimizing seed placement in pelvic seed implant brachytherapy. However, this task is challenging due to poor image contrast and needle adhesion. This paper presents a novel approach that reframes needle localization as a tip-handle detection and matching problem to overcome these difficulties. An anchor-free network, based on HRNet, is proposed to extract multi-scale features and accurately detect needle tips and handles by predicting their centers and orientations using decoupled branches for heatmap regression and polar angle prediction. To associate detected tips and handles into individual needles, a greedy matching and merging (GMM) method designed to solve the unbalanced assignment problem with constraints (UAP-C) is presented. The GMM method iteratively selects the most probable tip-handle pairs and merges them based on a distance metric to reconstruct 3D needle paths. Evaluated on a dataset of 100 patients, the proposed method demonstrates superior performance, achieving higher precision and F1 score compared to a segmentation-based method utilizing the nnUNet model,thereby offering a more robust and accurate solution for needle localization in complex clinical scenarios.

cs.CV↗
arXiv2026

Benchmarking and Evolving Reason-Reflect-Rectify for Reflective Visual Generation

Text-to-Image (T2I) models and Unified Multimodal Models (UMMs) have achieved remarkable progress in visual generation. However, their reliance on a single-pass generation paradigm limits their ability to handle complex prompts requiring iterative refinement. To enable multi-round Reflective Visual Generation (RVG), we formalize the Reason-Reflect-Rectify (R^3) loop as a core framework and introduce R^3-Bench, a benchmark of over 600 expert-annotated instances that quantifies iterative reasoning and rectification capabilities. Evaluation on R^3-Bench reveals a critical gap: while state-of-the-art models can identify generation errors, they fail to generate actionable rectification instructions. To bridge this gap, we propose R^3-Refiner, a dual-stage framework leveraging Group Relative Policy Optimization (GRPO) and a Hierarchical Reward Mechanism (HRM) to better align rectification with reflective reasoning. Experiments show that R^3-Refiner achieves significant improvements on R^3-Bench (+12.0% in Reflective Verdict Score, +9.0% in Rectification Score), and can be seamlessly integrated with various MLLMs to enhance the generation quality of different T2I models on GenEval++ and T2I-CompBench. Code is available at https://github.com/xiaomoguhz/R3-Bench.

cs.CV↗
arXiv2026

AgentSteerTTS: A Multi-Agent Closed-Loop Framework for Composite-Instruction Text-to-Speech

While existing text-to-speech (TTS) models exhibit high expressiveness, fine-grained control over composite instructions remains challenging due to the structural mismatch between discrete textual intents and continuous acoustic realizations. Inspired by human cognitive decoupling, we introduce AgentSteerTTS, a multi-agent closed-loop framework designed for intent-faithful expressive control of composite instructions. First, in our framework, an adversarial disentanglement agent mitigates speaker-emotion leakage by learning separable identity and emotion-prosody subspaces with leakage-suppressing regularization. Next, a Dual-Stream Anchoring Controller grounds abstract intents using a large-scale acoustic prototype library: a Retrieval Agent selects expressive anchors, while a Synthesis Agent fuses them into continuous control vectors via gated attention. Finally, a Fast-Slow Feedback Agent refines output intensity through latent gradient correction and resolves semantic-acoustic mismatches using high-level perceptual critique. Experiments on a composite-instruction benchmark and public test sets show that AgentSteerTTS yields consistent and significant improvements to the baselines, demonstrating the effectiveness of the proposed method.

cs.CV↗
arXiv2026

EXPEREPAIR: Dual-Memory Enhanced LLM-based Repository-Level Program Repair

Automatically repairing software issues remains a fundamental challenge at the intersection of software engineering and AI. Although recent advances in Large Language Models (LLMs) have demonstrated potential for repository-level repair tasks, current methods exhibit two notable limitations: (1) they often address issues in isolation, neglecting to incorporate insights from previously resolved issues, and (2) they rely on static, rigid prompting strategies that constrain their ability to generalize across diverse and evolving contexts. We propose ExpeRepair, a novel LLM-based program repair framework inspired by the dual-memory systems of human cognition, where episodic and semantic memory synergistically support learning and decision-making. Unlike existing methods, ExpeRepair continuously learns from historical repair experiences via dual-channel knowledge accumulation, enabling it to adaptively reuse past knowledge during inference. Specifically, ExpeRepair organizes prior repair knowledge into two complementary memories: an episodic memory that stores concrete repair demonstrations, and a semantic memory that encodes abstract, reflective insights. At inference time, ExpeRepair activates both memory systems by retrieving relevant demonstrations from episodic memory and recalling high-level repair insights from semantic memory. It further enhances adaptability through dynamic prompt composition, integrating both memory types to replace static prompts with context-aware, experience-driven prompts. We evaluate ExpeRepair on two benchmarks: SWE-Bench Lite and SWE-Bench Verified. Experimental results show that ExpeRepair achieves pass@1 scores of 60.3% and 74.6% on the two benchmarks, respectively, achieving the best performance among the evaluated open-source methods. We have open-sourced ExpeRepair at https://github.com/ExpeRepair/ExpeRepair.

cs.SE↗
arXiv2026

WebClipper: Efficient Evolution of Web Agents with Graph-based Trajectory Pruning

Deep Research systems based on web agents have shown strong potential in solving complex information-seeking tasks, yet their search efficiency remains underexplored. We observe that many state-of-the-art open-source web agents rely on long tool-call trajectories with cyclic reasoning loops and exploration of unproductive branches. To address this, we propose WebClipper, a framework that compresses web agent trajectories via graph-based pruning. Concretely, we model the agent's search process as a state graph and cast trajectory optimization as a minimum-necessary Directed Acyclic Graph (DAG) mining problem, yielding pruned trajectories that preserve essential reasoning while eliminating redundant steps. Continued training on these refined trajectories enables the agent to evolve toward more efficient search patterns and reduces tool-call rounds by about 20% while improving accuracy. Furthermore, we introduce a new metric called F-AE Score to measure the model's overall performance in balancing accuracy and efficiency. Experiments demonstrate that WebClipper compresses tool-call rounds under excellent performance, providing practical insight into balancing effectiveness and efficiency in web agent design.

cs.AI↗
arXiv2026

Velocity-Space 3D Asset Editing

Editing a 3D asset locally, modifying a target region while preserving the rest, is a fundamental requirement of native 3D editing. Existing methods enforce locality through mechanisms external to the generator, such as manual 3D masks, post-hoc voxel merging, or 2D multi-view lifting. None of them intervene where the corruption actually originates: inside the ODE sampler. For a rectified-flow generator to achieve faithful local editing, its velocity field should be strong over the target editing region while vanishing on preserved content. Yet a single velocity field can hardly satisfy both requirements simultaneously, leading to three problems: (i) identity leakage that keeps the edit signal non-zero on preserved regions; (ii) no dedicated edit-amplification channel, so strengthening the edit inevitably perturbs identity; and (iii) an identity drag at the geometry and material stages, where a global condition pulls every token toward the target. We propose VS3D (Velocity-Space 3D Asset editing}), an inversion-free, training-free, and mask-free framework that addresses each problem with a targeted intervention inside the sampler. VS3D integrates three complementary modules, each corresponding to a specific stage of the editing pipeline. Reconstruction-Anchored Source Injection (RASI) absorbs identity leakage by turning the unconditional embedding into a per-step, asset-specific anchor calibrated through source reconstruction. Partial-Mean Guidance (PMG) amplifies the edit signal by contrasting high- and low-quality subsample estimates of the velocity difference, active only where a consistent edit exists. Twin-Agreement Residual injection (TAR) lets the sampler decide token by token what to preserve at the geometry and material stages.

cs.GR↗
arXiv2026

OrchJail: Jailbreaking Tool-Calling Text-to-Image Agents by Orchestration-Guided Fuzzing

Tool-calling text-to-image (T2I) agents can plan and execute multi-step tool chains to accomplish complex generation and editing queries. However, this capability introduces a new safety attack surface: harmful outputs may arise from tool orchestration, where individually benign steps combine into unsafe results, making prompt-only jailbreak techniques insufficient. We present OrchJail, an orchestration-guided fuzzing framework for jailbreaking tool-calling T2I agents. Its core idea is to exploit high-risk tool-orchestration patterns: by learning from successful jailbreak tool-calling traces and their causal relationships to prompt wording, OrchJail directly guides the fuzzing search toward prompts that are more likely to trigger unsafe multi-step tool behaviors, rather than relying on surface-level textual perturbations. Extensive experiments demonstrate that OrchJail improves jailbreak effectiveness and efficiency across representative toolcalling T2I agents, achieving higher attack success rates, better image fidelity, and lower query costs, while remaining robust against common jailbreak defenses. Our work highlights tool orchestration as a critical, previously unexplored attack surface and provides a novel framework for uncovering safety risks in T2I agents.

cs.MA↗