Searcharxiv⌕ Search

arXiv subjects

Yuhan Chen

Publications and source records attributed to Yuhan Chen.

At least 19 recordsLinked to original sources
arXiv2026

GSAR: Goal-State-Anchor Rewards for Mobile GUI Agents with Self-Evolving Data Synthesis

Vision-Language Models (VLMs) based GUI agents stand to benefit significantly from online reinforcement learning (RL). However, their training is bottlenecked by two fundamental issues: current data synthesis methods for GUI Agents rely on specific environments and struggle to generate diverse data, while existing evaluators either suffer from limited scalability or provide inaccurate and unreliable reward signals. To overcome these challenges, we introduce GSAR (Goal-State-Anchor Reward), a RL reward framework that supports scalable task generation and delivers reliable reward signals for stable and efficient policy optimization. Our approach features self-evolving data synthesis, which produces multiple environments through task execution and generates diverse tasks and goal states. Complementing this, a state-anchor mechanism automatically annotates task-relevant UI elements in successful goal states as reference anchors. During RL training, these reference anchors provide accurate, scalable reward signals that substantially enhance efficiency. Extensive evaluations demonstrate that our framework achieves over 90% accuracy on offline trajectory verification and performs closest to rule-based methods. Furthermore, agents trained using our reward framework exhibit strong performance on both AndroidWorld and our constructed benchmark, establishing a scalable approach for GUI agent training.

cs.AI↗
arXiv2026

Beyond Compression Ratio: High-Throughput KV Compression for Disaggregated LLM Serving

KV cache compression reduces the communication volume of disaggregated large language model serving, but its practical benefit depends not only on compression ratio and model quality, but also on how efficiently the compression pipeline executes. This requirement becomes increasingly important as modern serving systems adopt high-bandwidth interconnects, where inefficient compression can offset a substantial portion of the saved communication time. This paper presents \system, a high-throughput KV cache compression data path built on top of \kvserve. \system improves compression execution in two complementary ways. First, it fuses the distribution transformation and low-bit quantization stages into a KV-aware \tilelang kernel, eliminating intermediate tensor materialization and redundant memory accesses. Second, it replaces the original lossless compression backend with an open, extensible, and searchable backend based on \lc, enabling high-performance compression operator pipelines. The fused kernel achieves up to $5.9\times$ speedup, while the new lossless backend increases compression throughput from below 90~GB/s to over 500~GB/s. Across the evaluated workloads, \system reduces job completion time by up to $2.04\times$. End-to-end evaluation shows that these optimizations allow KV compression to retain meaningful communication benefits under modern high-bandwidth network.

cs.DC↗
arXiv2026

ProAS: A Progress-Aware Advantage Shaping Framework for Multi-Turn Agentic Reinforcement Learning

Reinforcement learning (RL) has emerged as the primary paradigm for optimizing Large Language Model (LLM) agents, yet efficient critic-free algorithms (e.g., GRPO) suffer from a fundamental uniform credit assignment dilemma that broadcasts trajectory-level advantages equally to all actions. Our analysis reveals that this uniformity inherently reinforces structurally unreasonable actions in preferred trajectories and ignores each action's downstream deviations in unpreferred ones. To resolve this without introducing critic overhead, we propose Progress-Aware Advantage Shaping (ProAS), a plug-and-play framework that leverages a dynamically evolving Agentic Reference Graph (ARG) to achieve fine-grained credit assignment via Masking Unreasonable Incentives (MUI) and Asymmetric Penalty Redistribution (APR). Extensive experiments across Qwen2.5-1.5B-Instruct and Qwen2.5-7B-Instruct demonstrate that ProAS universally elevates mainstream group-based baselines to new state-of-the-art performance, achieving peak success rates of 91.9% and 94.3% on ALFWorld, alongside 66.7% and 70.8% on WebShop, respectively. Furthermore, it accelerates training convergence by up to $2.44\times$ while maintaining strong robustness against imperfect or distilled ARGs. Codes: https://github.com/LINs-lab/ProAS

cs.LG↗
arXiv2026

CaliMem: Search-Efficient Data Agent via Calibrated Memory

Data curation is a first-order design choice in machine learning, yet tailoring training data to a downstream model and task remains costly in labor, expertise, and compute. Datasets are therefore often processed through fixed pipelines and reused across applications. We show that this assumption is fragile: across 55 dataset--model tasks spanning biological domains, the same curation recipe can help one model while hurting another. Effective curation must adapt to the downstream setting, motivating automated search over data recipes. We first build an agentic system that searches for curation policies at deployment, improving data utility via \textit{test-time curation (TTC)}. Although achieving performance gains, its search is driven largely by recent feedback and remains local. Adding persistent memory helps the agent retain evidence across iterations and reveals a deeper problem: it forms claims that extend beyond the conditions in which they were observed, sometimes abandoning an entire strategy after a single unsuccessful trial. To address this problem, we introduce CaliMem, a calibrated-memory agentic loop that requires the agent to predict the outcome of each proposed policy before executing it. Comparing this prediction with the observed result gives the agent a \textit{reality check} on the claims behind its proposal. It uses the prediction error to retain supported claims, revise mistaken ones, and specify the conditions under which a conclusion holds. This makes the agent's beliefs testable, reduces overgeneralization, and supports more effective long-horizon exploration. Across biological and LLM post-training tasks, CaliMem consistently outperforms fixed curation baselines and published methods, and exceeds full-data training by up to 19.4\%. Code is open-sourced.

cs.AI↗
arXiv2026

Heuristics are Not Enough: Calibrated and Corrected Caching for Video Diffusion Acceleration

Feature caching has become a mainstream training-free strategy for accelerating video diffusion models by reusing intermediate outputs across denoising steps. However, existing adaptive methods rely on heuristic difference measures misaligned with actual generation quality, employ a single global threshold that ignores heterogeneous phase dynamics, and never validate their budget estimates against true approximation error, allowing systematic bias to propagate uncorrected. We propose \textbf{AlignCache}, a closed-loop caching framework that directly aligns scheduling decisions with their true influence on the final generated video through three tightly coupled components: (1) phase-adaptive budget routing that partitions the denoising trajectory into structurally distinct phases, each governed by an independent threshold and a staleness-aware guard; (2) closed-loop threshold calibration that exploits counterfactual error measurement at the classifier-free guided output level to adjust decision boundaries online via an EMA feedback loop; and (3) momentum-augmented draft synthesis combining direction-magnitude gated residual extrapolation with anchor-guided unconditional correction. By transforming scheduling from a static risk accumulator into a feedback-driven, self-correcting controller, AlignCache consistently achieves superior acceleration-quality trade-offs across state-of-the-art video generation models. Our code is available at https://github.com/jiangjyjy/AlignCache.

cs.CV↗
arXiv2026

Hybrid Latent Attention for Looped Language Models

Looped language models apply the same stack of layers T times to each token, which deepens the model without adding parameters but multiplies its key-value (KV) cache by T. The larger cache limits how many sequences a GPU can decode at once and slows each decoding step, which reads the whole cache. We propose Hybrid Latent Attention (HLA), which keeps exact keys and values within a sliding window of W recent tokens and stores each older token as a compact latent that the query of each loop reads directly, without reconstructing keys and values. We uptrain HLA on Ouro looped models (T=4) with 1.4B and 2.6B parameters, keeping the pretrained weights frozen and training only the added parameters to reproduce the original attention. The cache shrinks by 10.7x per token, fitting 4.0-8.8x as many concurrent sequences per GPU, and decoding throughput improves by 2.5x at 1K-token contexts and by up to 7.4x at 16K. HLA retains over 97% of the original accuracy on math, knowledge and reasoning benchmarks, and 96-100% on long-context retrieval up to 16K tokens. After supervised fine-tuning, it performs on par with the fine-tuned original model on competition-level math.

cs.CL↗
arXiv2026

RAVEL: Asynchronous Rolling Inference for Flow-Based Vision-Language-Action Models

Flow-based vision-language-action (VLA) models are highly effective for generalist robot manipulation, yet their reliance on computationally expensive VLM encoding and multi-step iterative action generation imposes a significant latency bottleneck. The resulting inference latency makes it difficult for robots to respond quickly, especially in dynamic environments. We address this limitation with RAVEL (Rolling Asynchronous VLA Enabling Low-Latency Control), an asynchronous inference framework that addresses the computational bottlenecks of both the VLM backbone and the action expert. To reduce the delay from multi-step action denoising, RAVEL allows near-term actions to be executed after a single denoising step by carrying partially denoised future actions forward in a rolling buffer. To avoid blocking on slow VLM encoding, RAVEL decouples VLM encoding from rolling action generation, allowing the action expert to operate continuously using the latest available VLM context, while a lightweight Fast Observation Pathway (FOP) directly conditions the action expert on current observations. Across simulated and real-world manipulation tasks, RAVEL consistently achieves substantially lower response latency while maintaining the task capability of the underlying VLA, enabling high-frequency and responsive closed-loop control.

cs.RO↗
arXiv2026

Can Frozen Hyperspherical Features Guide the Selection of Pseudo Masks?

Foundation segmenters such as SAM return several plausible masks for an unlabeled image, and a student trained on the wrong one inherits its errors. Choosing among them means querying a second large model or fitting a quality head to annotated masks. We show that a candidate can be judged by what it does to a frozen self-supervised backbone's features. Normalized DINOv2 patch features lie on a hypersphere, and a candidate mask splits that sphere in two. Based on this reading, we introduce SphereTrust, which scores each candidate by three properties of the split, the angular contrast between the two sides, the coverage of the foreground's appearance modes, and contact with the image frame, one for each of three common ways a mask fails, and ranks a pool in 0.55 s per image from the frozen features alone. On eight SAM and SAM3 candidate pools spanning camouflaged, salient, and dichotomous segmentation and camouflage under low light, SphereTrust exceeds the strongest evaluated external baseline on six pools by 1.7 to 9.3 percentage points in mean selected Dice. These comparisons include published selection rules and explicitly labeled adaptations of DSS and UCOD-MKD. On the two prompted camouflage pools, its mean selected Dice is within 0.1 percentage points of the candidate-derived DSS adaptation, with a lower catastrophic-error rate. Which cue carries the signal depends on the candidate pool. The same sphere also supports training. The leading candidates enter as a candidate set with their scores as priors, prototypes reorder them, and a cross-fitted second round completes the labels, raising weighted F by 4.5, 2.3, and 5.5 points over fixed-label training on the three MLLM anchor pools, with students competitive with published unsupervised methods on nineteen test sets.

cs.CV↗
arXiv2026

No Free Checker: A Survey of Verifiers for Robot Policies

A verifier for robot policies reads a candidate behavior and returns a score for how well it did, used both to evaluate vision-language-action policies and to train them. Verifiers range from success detectors and reward models to runtime monitors, safety filters, and temporal-logic specifications. We survey roughly 150 verifiers and compare them along two properties. Availability is how much a verdict costs, how early in a rollout the verdict arrives, and how often a verdict can be asked for. Availability rises as verdicts get cheaper, earlier, and denser. Credibility is how much a high score tells us about the task. Credibility falls as the judgment becomes gameable and self-serving. We group the verifiers by who supplies the judgment: human verifiers, rule-based and formal verifiers, learned and pretrained verifiers, and model-intrinsic verifiers. Across the four families, we find that credibility falls as availability rises. Regardless of who supplies the judgment, there is no free checker. We then examine what validates a verifier itself, and how much a high score tells us. Three measures appear in the literature: agreement with human labels, the performance of the policy it trains, and behavior under reward hacking. We close with nine metrics that make a verifier claim checkable, and coordinates for the verifiers still to be built.

cs.RO↗
arXiv2026

Hybrid Event Frame Sensors: Modeling, Calibration, and Simulation

Hybrid event-frame sensors integrate an Event Vision Sensor (EVS) and an Active Pixel Sensor (APS) within a single chip, combining the high dynamic range and low latency of the EVS with the rich spatial intensity information from the APS. While this tight integration offers compact and temporally precise imaging, the complex circuit architecture introduces nontrivial noise patterns that remain poorly understood and unmodeled. In this work, we present the first unified statistics-based imaging noise model that jointly describes the noise behavior of APS and EVS pixels. Our formulation explicitly incorporates photon shot noise, dark current noise, fixed-pattern noise, and quantization noise, and links EVS noise to illumination level and dark current. Based on this formulation, we further develop a calibration pipeline to estimate noise parameters from real data and provide a detailed analysis of both APS and EVS noise behaviors. Finally, we propose H-ESIM, a statistically grounded simulator that generates RAW frames and events under realistic jointly calibrated noise statistics. Experiments on two hybrid sensors validate our model across multiple imaging tasks, including video frame interpolation and deblurring, demonstrating strong transfer from simulation to real data.

cs.CV↗
arXiv2026

Belief-Calibrated Optimization: An Explicit World Model for Agentic Optimization

The performance of an LLM agent depends on the scaffold around a frozen model. A common way to improve that scaffold is to use a coding agent as an optimizer: it reads current scores and traces and iteratively edits the source, producing a new candidate each round. Each edit is chosen according to a belief about how the environment will respond: what went wrong, and which change should help. That belief is typically implicit. It lives in the coding agent's reasoning on the current call, or remains latent in its parameters, rather than as something written down. Later calls therefore see scores and traces, but they do not use that belief. We introduce Belief-Calibrated Optimization (BCO), a method that writes that belief down as a persistent in-context document and continually revises that document as new candidates are evaluated. The resulting document is a world model: the current account of how the environment responds to edits. Added to an otherwise standard loop, BCO reaches a higher train passrate than a matched control that lacks only the world model, on five benchmarks spanning memory QA, tool-use QA, code-as-action app agents, and terminal agents. The gap remains on every held-out split, which is not used to select the candidate. After a target-model swap, in which the frozen model is replaced and the scaffold is not, the selected BCO scaffold leads on the tasks we test, except where context-window overruns leave it unfinished. An offline ablation then asks whether that gap comes from what the world model says. A fresh predictor given the accumulated document forecasts how the environment will respond more accurately than predictors given either no document or a same-form copy whose content has been falsified. The comparison indicates that the document carries reusable information in its content, not only in its form.

cs.AI↗
arXiv2026

UltraFast-LiNET: Light-weight multi-scale shift convolutional network for real-time low-light image enhancement

Addressing the urgent need for high-performance real-time low-light image enhancement on resource-constrained edge devices in low-illumination scenarios such as nighttime and tunnels, this paper presents UltraFast-LiNET, an ultra-lightweight network designed for extreme efficiency. The proposed network introduces Dynamic Shift Convolution, DSConv, a highly compact operation with only 12 learnable parameters. By using DSConv to construct a Multi-Scale Shift Residual Block, MSRB, UltraFast-LiNET effectively enlarges the receptive field while maintaining an exceptionally small parameter count. To alleviate gradient instability during the training of lightweight models, we further propose a multi-level gradient-aware loss function that strengthens supervision signals across different feature levels. UltraFast-LiNET supports flexible parameter configurations, enabling the model size to be reduced to as few as 36 learnable parameters. Remarkably, with only 180 learnable parameters, UltraFast-LiNET achieves a PSNR of 19.81 dB on the LOL dataset, surpassing the current state-of-the-art by 0.08 dB. Experimental results demonstrate that the proposed method achieves an effective balance between millisecond-level real-time processing and superior enhancement quality under extreme computational constraints. These results highlight its chip-level computational efficiency and broad prospects for edge-side deployment.

cs.CV↗
arXiv2026

Triton for MTIA: Bridging the Programming Model Gaps for Custom AI Accelerators

The rapid growth in machine learning workloads has fueled the proliferation of custom accelerator architectures. Designed from the ground up, these accelerators often expose programming models that are distinct from GPUs. While hyperscalers and AI chip startups continue to innovate in this space, achieving broad operator coverage to support diverse models remains a major challenge. Additionally, an easy-to-use, high-level kernel programming language is important for rapid iteration of models and kernels. Triton, together with TorchInductor, addresses these issues on GPUs, but its viability on accelerators with different programming models has yet to be established. In this work, we present the first production-scale application of Triton on a custom ML accelerator, MTIA-2i, developed by Meta. To support MTIA-2i, we develop a new compiler backend that targets it, introduce enhancements to TorchInductor code generation, and propose minimal language extensions that expose MTIA-specific architectural features. We demonstrate that Triton-MTIA kernels achieve performance competitive with expert-tuned C++ implementations. Leveraging these development efficiency gains, we successfully deployed manually written and Inductor-generated Triton kernels in production across approximately 60 different model types, accounting for 50% of layers and 47% of non-GEMM execution time for these models. Our results provide compelling evidence that DSLs like Triton can bridge the programming model gaps between ML frameworks, kernels, and custom accelerators, enabling rapid innovation and efficient deployment at scale.

cs.PL↗
arXiv2026

CT Open: An Open-Access, Uncontaminated, Live Platform for the Open Challenge of Clinical Trial Outcome Prediction

Scientists have long sought to accurately predict outcomes of real-world events before they happen. Can AI systems do so more reliably? We study this question through clinical trial outcome prediction, a high-stakes open challenge even for domain experts. We introduce CT Open, an open-access, live platform that will run four challenge every year. Anyone can submit predictions for each challenge. CT Open evaluates those submissions on trials whose outcomes were not yet public at the time of submission but were made public afterwards. Determining if a trial's outcome is public on the internet before a certain date is surprisingly difficult. Outcomes posted on official registries may lag behind by years, while the first mention may appear in obscure articles. To address this, we propose a novel, fully automated decontamination pipeline that uses iterative LLM-powered web search to identify the earliest mention of trial outcomes. We validate the pipeline's quality and accuracy by human expert's annotations. Since CT Open's pipeline ensures that every evaluated trial had no publicly reported outcome when the prediction was made, it allows participants to use any methodology and any data source. In this paper, we release a training set and two time-stamped test benchmarks, Winter 2025 and Summer 2025. We believe CT Open can serve as a central hub for advancing AI research on forecasting real-world outcomes before they occur, while also informing biomedical research and improving clinical trial design. CT Open Platform is hosted at $\href{https://ct-open.net/}{https://ct-open.net/}$

cs.AI↗
arXiv2026

DeepImagine: Clinical Trial Outcome Prediction via Stepwise Local Counterfactual Imaginations

Predicting the outcomes of prospective clinical trials remains a major challenge. Clinical trial outcomes result from complex interactions among experimental factors such as drug interventions, participant demographics, and protocols. Here, we introduce DeepImagine, a framework that predicts target trial outcomes through stepwise counterfactual imagination anchored on historical trials with observed results. Starting from a relevant historical trial, DeepImagine sequentially modifies one differing experimental factor at a time. With each step a large language model (LLM) is posed a local counterfactual: how would the current imagined outcome change with this single perturbation? The updated result is carried forward as the input to the next step, until the historical configuration exactly matches the target, yielding the final prediction. Empirically, DeepImagine consistently outperforms direct one-step prediction across several off-the-shelf LLMs, with further gains when multiple imagination pathways, initiated from different historical anchors, are aggregated. We also construct natural counterfactuals augmented with synthetic reasoning traces and train a family of specialized language models, each dedicated to learning one factor's local counterfactual transition. Integrating these learned local operators into DeepImagine yields substantial improvements over general-purpose LLM baselines. Our findings position stepwise counterfactual imagination, distinct from both correlational prediction and explicit structural causal modeling, as a promising direction for clinical trial outcome prediction. All code, training scripts, and evaluation scripts are available at \href{https://github.com/deepimagine-counterfactual/DeepImagine}{https://github.com/deepimagine-counterfactual/DeepImagine}.

cs.CL↗
arXiv2026

MIND: Multimodal Intent-Driven Network via Diffusion Transformers for Medical Image Fusion

Medical image fusion aims to integrate complementary information from diverse imaging modalities to support clinical diagnosis. Existing methods typically apply uniform fusion rules globally, lacking a deep understanding of diagnostic intents and pathological structures. To address these limitations, we propose MIND, a Multimodal Intent-Driven Network via Diffusion Transformers (DiTs) for medical image fusion. Specifically, we utilize BioMedGPT to generate intent-driven fusion texts from source images, guiding the fusion process with pathology-aware diagnostic intents. To combat the loss of 2D spatial continuity caused by 1D sequence flattening in DiTs, we design a Multi-scale Latent Adapter. This module explicitly extracts source image features before serialization, injecting them into the network via strict dimensional alignment to effectively supplement image features. To resolve the semantic shift caused by decoupling image outputs from diagnostic intents, we design a medical semantic consistency loss. This loss ensures deep semantic locking between fused images and fusion texts while maintaining the stability of the underlying physical manifold reconstruction. Comprehensive experiments on the Harvard, BraTS, and GFP datasets reveal that MIND delivers superior fusion quality, significantly improves downstream brain tumor segmentation accuracy, and enables flexible interactive fusion, holding significant promise for intent-driven intelligent clinical decision support systems.

cs.CV↗
arXiv2026

Full-Path Nonlinear Modeling of Microwave Power Transmission Through Ionospheric Plasma for Space Solar Power Station

Space Solar Power Station (SSPS) concepts rely on gigawatt-class microwave beams to carry orbital solar energy through the ionosphere, where the beam and the plasma form a coupled nonlinear system: the field heats electrons, the heating alters the collision frequency and plasma density, and the modified medium in turn reshapes the field. To our knowledge, this work is the first study to quantify this two-way interaction between microwave power transmission and the ionospheric plasma environment through full-path nonlinear modeling. The 340 km path from 400 km to 60 km altitude is reconstructed by 34 cascaded two-dimensional axisymmetric finite-element full-wave segments with complex-field transfer, using International Reference Ionosphere (IRI) electron-density and NRLMSISE-00 neutral-atmosphere inputs. A Shallow Neural Network (SNN) surrogate replaces the implicit electron energy balance with an explicit closure that maps altitude and local field magnitude to electron temperature and effective collision frequency, enabling stable nonlinear iteration. For 1 GW beams at 2.45 GHz and 5.8 GHz, the volume-integrated Ohmic deposition is 29.4 kW and 5.11 kW, respectively -- fractional losses of order $10^{-5}$ -- and the ratio between the two bands follows the $ω^{-2}$ scaling of collisional absorption. The deposition concentrates near 95 km altitude, where the product of electron density and collision frequency peaks, whereas the electron-temperature perturbation (up to 3815 K) maximizes in the F region, where cooling is weakest; ponderomotive density depletion remains below 0.02\%. The ionosphere is therefore effectively transparent to the SSPS power budget but not to the beam phase: localized heating and refractive perturbation accumulate phase-front distortion relevant to phased-array beam control, rectenna phase compensation, and environmental assessment.

physics.plasm-ph↗
arXiv2026

Think Before You Comfort: Reflective Cognitive Alignment for Protocol-Grounded Elderly Stimulation Agents

Cognitive Stimulation Therapy (CST) offers non-pharmacological support for elders with cognitive impairment, yet scalability remains constrained by reliance on trained facilitators and severe data scarcity, particularly for privacy-sensitive, low-resource languages such as Cantonese. While Large Language Models (LLMs) show promise for automated companionship, they often struggle to balance empathetic engagement with adherence to cognitive stimulation guidelines. We propose a framework addressing these challenges along two complementary axes. First, STaR-CS (Style-Transfer and Role-Conditioned Cognitive Stimulation) synthesizes multi-party dialogues through facilitator style modeling and structured skeleton extraction, mitigating data barriers. Building upon this corpus, the Reflective Cognitive Alignment (RCA) framework models stimulation interactions as a sequential decision process, integrating Protocol-Constrained Chain-of-Cognition (PC-CoC) for structured reasoning and Inference-Time Value Alignment (IVA) for principled response selection based on safety and engagement goals. Evaluations across six backbone LLMs and two independent judges show that RCA consistently improves protocol adherence, safety, and group facilitation over standard prompting baselines. Our code is available at https://github.com/jiangjyjy/RCA_Agent.

cs.CL↗