Search arXiv⌕ Search

arXiv subjects

Tao Yang

Publications and source records attributed to Tao Yang.

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

Real-World Blind Super-Resolution via Feature Matching with Implicit High-Resolution Priors

A key challenge of real-world image super-resolution (SR) is to recover the missing details in low-resolution (LR) images with complex unknown degradations (e.g., downsampling, noise and compression). Most previous works restore such missing details in the image space. To cope with the high diversity of natural images, they either rely on the unstable GANs that are difficult to train and prone to artifacts, or resort to explicit references from high-resolution (HR) images that are usually unavailable. In this work, we propose Feature Matching SR (FeMaSR), which restores realistic HR images in a much more compact feature space. Unlike image-space methods, our FeMaSR restores HR images by matching distorted LR image features to their distortion-free HR counterparts in our pretrained HR priors, and decoding the matched features to obtain realistic HR images. Specifically, our HR priors contain a discrete feature codebook and its associated decoder, which are pretrained on HR images with a Vector Quantized Generative Adversarial Network (VQGAN). Notably, we incorporate a novel semantic regularization in VQGAN to improve the quality of reconstructed images. For the feature matching, we first extract LR features with an LR encoder consisting of several Swin Transformer blocks and then follow a simple nearest neighbour strategy to match them with the pretrained codebook. In particular, we equip the LR encoder with residual shortcut connections to the decoder, which is critical to the optimization of feature matching loss and also helps to complement the possible feature matching errors. Experimental results show that our approach produces more realistic HR images than previous methods. Codes are released at https://github.com/chaofengc/FeMaSR.

cs.CV↗
arXiv2026

Experience Graphs: The Data Foundation for Self-Improving Agents

The database community has repeatedly advanced the state of the art by recognizing that new workloads demand new system architectures. We argue that long-horizon agentic tasks -- code generation, scientific discovery, hardware design -- are such a workload. These agents explore: they generate artifacts, execute tools, observe failures, branch, and repair over hundreds of steps. This search produces a structured object we call an experience graph: executable artifacts, tool outputs, rewards, sibling comparisons, and causal lineage. Yet existing agent frameworks treat this experience as disposable state -- JSON checkpoints and session logs that cannot be recovered after a crash, queried across users, or materialized into training data. We propose Trellis: a data foundation that treats the experience graph as first-class, governed, queryable database state. The core insight is that search over experience graphs is a database access pattern. Frontier selection is a query, cross-session reuse is vector-seeded graph retrieval, training-data extraction is a materialized view, and reconstructing what an agent knew at any past step is a time-travel query. When the database owns the experience graph, agents become stateless compute, and crash recovery, horizontal scaling, and a closed-loop training flywheel emerge as architectural byproducts. We ground the design in KernelEvolve, a production accelerator-kernel optimizer at Meta, where cross-session reuse reaches a target speedup roughly 10x faster at 52% lower token cost. More broadly, Trellis turns inference-time search from disposable computation into a durable institutional asset: logs made databases reliable; experience graphs may make agents cumulative.

cs.DB↗
arXiv2026

PAMAE: Phase-Aware-MoE Action Experts Towards Reliable Flow-Matching Vision-Language-Action Policies

Reliable action generation for multi-stage robotic manipulation remains challenging for Vision-Language-Action (VLA) models. While existing flow-matching VLA policies offer strong multimodal grounding and generalization, they typically employ a single shared action expert, limiting their ability to capture phase-specific control patterns across distinct execution stages. We propose a plug-and-play Phase-Aware Mixture-of-Experts Action Module (PAMAE), as a step towards more reliable phase-consistent action generation. PAMAE replaces the original flow-matching action expert with a sparse expert mixture while preserving the pretrained VLA backbone. PAMAE introduces a phase-aware router that leverages execution-phase cues to allocate action generation across experts, supported by a lightweight phase prediction head and a routing alignment objective. To stabilize specialization, we adopt a two-stage training scheme that first warms up the expert module under the standard flow-matching loss and then optimizes phase-consistent routing under auxiliary supervision. On multi-stage manipulation simulation tasks, PAMAE improves task success by up to \textbf{9.2\%} over strong VLA baselines. Further ablations show that both phase-supervised routing and staged optimization are essential for the observed gains. Our results highlight phase-consistent expert allocation as an effective mechanism for improving the reliability and action quality of flow-matching VLA policies.

cs.RO↗
arXiv2026

PhysReflect-VLA: Physical Feasibility and Self-Reflective Regulation for Reliable Vision-Language-Action Policies

Long-horizon robotic manipulation is highly sensitive to physically infeasible transitions, contact-induced disturbances, and the lack of effective self-correction during execution. Although Vision-Language-Action (VLA) models provide strong task grounding through multimodal learning, they typically generate actions in a feed-forward manner without explicitly checking physical feasibility or diagnosing execution errors online. We present PhysReflect-VLA, a plug-and-play execution-time reliability framework that augments VLA policies with physical feasibility evaluation and structured self-reflection in a closed-loop control pipeline. A Feasibility Operator evaluates whether candidate actions induce dynamically consistent state transitions; an Action Explanation Operator verifies transition coherence; and an LLM-based Reflection Module analyzes state discrepancies to generate corrective guidance for subsequent actions. A two-stage training procedure stabilizes feasibility modeling and integrates reflection into the control loop. Experiments on multi-stage, contact-rich real-world manipulation tasks show consistent improvements in stage-wise stability and overall task success compared with representative VLA baselines with an average gain of 5.4\%. Ablation results further indicate that feasibility checking and reflection-based correction both contribute to improved execution robustness. These results highlight the importance of embedding physical consistency checks and online self-reflection for reliable long-horizon robotic manipulation.

cs.RO↗
arXiv2026

A 3D Isovist World Model -- Revealing a City's Unseen Geometry and Its Emergent Cross-City Signature

Embodied agents that navigate cities rely on world models that predict how their surroundings will change as they move. But for navigation, what matters is not what the buildings look like; it is where the agent can go. Most world models nonetheless predict appearance, learning how a scene looks rather than the space an agent can move through. Those that do target geometry, such as bird's-eye-view occupancy grids, flatten the three-dimensional environment onto a ground plane, discarding the above-ground and multi-level structure that shapes real navigation. What is missing is a predictive target that captures the navigable geometry an agent actually traverses, without photometric entanglement and without collapsing the third dimension. Our key idea is to model the open volume between buildings, the negative space, encoded as a 3D isovist: a spherical visibility-depth map recording the distance to the nearest surface in every direction. We introduce an embodied world model that predicts the next isovist from a short history of past isovists and a movement action. The prediction is formulated as a depth residual so the decoder inherits sharp building edges, trained with self-rollout scheduled sampling to keep corrupted context on the geometry manifold, and equipped with a persistent latent bird's-eye-view spatial map for cross-path consistency. Our central finding is emergent and unexpected: a single city-blind model trained on Manhattan and Paris develops a cross-city spatial signature, with city identity linearly decodable from its temporal latents far above single-frame baselines, so the signature lives in the learned dynamics rather than in appearance. The representation is lightweight, interpretable, and reproducible, offering a geometric substrate for spatial reasoning in embodied AI, robotics, and urban analysis, released with an open dataset and pipeline.

cs.RO↗
arXiv2026

Parameter estimation of eccentric massive black hole binaries with LISA and its cosmological implications

Future space-based gravitational wave (GW) observatories such as LISA will detect massive black hole binaries (MBHBs), which are expected to be accompanied by electromagnetic counterparts, thereby providing bright standard sirens for cosmology. The orbital eccentricity of MBHBs can significantly improve the parameter estimation of GWs because the multiple harmonics induced by eccentricity provide additional information and help break down the degeneracies among waveform parameters. In this paper, we use the EccentricFD waveform and construct 5-year GW event catalogs for LISA under three population models (popIII, Q3d and Q3nod). For the three models, we find that an initial eccentricity of $e_0=0.4$ at $10^{-4}$ Hz yields improvements in sky localization and distance inference by a factor of $\mathcal{O}(10)$ in the best cases. As a consequence, the average number of bright sirens increases substantially: from 8 to 11 (PopIII), 6 to 12 (Q3d) and 13 to 24 (Q3nod). This increase in event number, together with enhanced localization and distance inference, leads to tighter cosmological constraints. In the $Λ$CDM model, for instance, the relative uncertainty on $H_0$ is reduced from $8.17\%$ to $4.35\%$ for the Q3d model, corresponding to an improvement of approximately $47\%$. We also investigate the improvement in constraints on the dark energy equation of state and modified GW propagation when combining bright sirens with the latest cosmic microwave background data. These results demonstrate that eccentricity is a remarkably significant feature in GW detection and parameter estimation, enabling more accurate measurements of the Universe with future space-based observatories.

astro-ph.IM↗
arXiv2026

KnapSpec: Self-Speculative Decoding via Adaptive Layer Selection as a Knapsack Problem

Self-speculative decoding (SSD) accelerates LLM inference by skipping layers to create an efficient draft model, yet existing methods often rely on static heuristics that ignore the dynamic computational overhead of attention in long-context scenarios. We propose KnapSpec, a training-free framework that reformulates draft model selection as a knapsack problem to maximize tokens-per-time throughput. By decoupling Attention and MLP layers and modeling their hardware-specific latencies as functions of context length, KnapSpec adaptively identifies optimal draft configurations on the fly via a parallel dynamic programming algorithm. Furthermore, we provide the first rigorous theoretical analysis establishing cosine similarity between hidden states as a mathematically sound proxy for the token acceptance rate. This foundation allows our method to maintain high drafting faithfulness while navigating the shifting bottlenecks of real-world hardware. Our experiments on Qwen3 and Llama3 demonstrate that KnapSpec consistently outperforms state-of-the-art SSD baselines, achieving up to 1.47x wall-clock speedup across various benchmarks. Our plug-and-play approach ensures high-speed inference for long sequences without requiring additional training or compromising the target model's output distribution.

cs.LG↗
arXiv2026

Stabilizing Policy Optimization via Logits Convexity

While reinforcement learning (RL) has been central to the recent success of large language models (LLMs), RL optimization is notoriously unstable, especially when compared to supervised fine-tuning (SFT). In this work, we investigate the stability gap between SFT and RL from a gradient-based perspective, and show that the convexity of the SFT loss with respect to model logits plays a key role in enabling stable training. Our theoretical analysis demonstrates that this property induces favorable gradient directionality during optimization. In contrast, Proximal Policy Optimization (PPO), a widely adopted policy gradient algorithm utilizing a clipped surrogate objective, lacks this stabilizing property. Motivated by this observation, we propose Logits Convex Optimization (LCO), a simple yet effective policy optimization framework that aligns the learned policy with an optimal target derived from the original RL objective, thereby emulating the stabilizing effects of logits-level convexity. Extensive experiments across multiple model families show that our LCO framework consistently improves training stability and outperforms conventional RL methods on a broad range of benchmarks.

cs.LG↗
arXiv2026

Fundamental Physics and Cosmology with TianQin

The exploration of the surrounding world and the universe is an important theme in the legacy of humankind. The detection of gravitational waves is adding a new dimension to this grand effort. What are the fundamental physical laws governing the dynamics of the universe? What is the fundamental composition of the universe? How has the universe evolved in the past and how will it evolve in the future? These are the basic questions that press for answers. The space-based gravitational wave detector TianQin will tune in to gravitational waves in the millihertz frequency range ($10^{-4} \sim 1$ Hz, to be specific), opening a new gravitational wave spectrum window to explore many of the previously hidden sectors of the universe. TianQin will discover many astrophysical systems, populating the universe at different redshifts: some will be of new types that have never been detected before, some will have very high signal-to-noise ratios, and some will have very high parameter estimation precision. The plethora of information collected will bring us to new fronts on which to search for the breaking points of general relativity, the possible violation of established physical laws, the signature of possible new gravitational physics and new fundamental fields, and to improve our knowledge on the expansion history of the universe. In this white paper, we highlight the advances that TianQin can bring to fundamental physics and cosmology.

gr-qc↗
arXiv2026

Beyond Exposure: Optimizing Ranking Fairness with Non-linear Time-Income Functions

Ranking systems in web search and recommendation allocate attention among items and providers, and therefore need to balance relevance-based effectiveness with provider fairness. Existing fair-ranking methods commonly focus on exposure fairness, where cumulative exposure is allocated in proportion to item merit. However, exposure is often only an intermediate signal: the actual utility received by a provider may depend on context-dependent conversion from exposure to income, such as clicks, purchases, or advertising value. This paper studies fair ranking under context-dependent provider utility, which we refer to as income. We formalize income fairness by requiring cumulative provider income to be proportional to relevance, and define an income-unfairness metric based on this proportionality condition. We then propose DIDRF, a Dynamic-Income-Derivative-aware Ranking Fairness algorithm for income-fair ranking. DIDRF uses the quadratic structure of income-fairness violations to derive a state-aware scoring rule that jointly considers ranking effectiveness and the marginal effect of each ranking decision on cumulative income fairness. Experiments on standard learning-to-rank datasets with log-calibrated semi-synthetic income environments based on advertising and e-commerce logs show that DIDRF consistently improves income fairness over representative fair-ranking baselines while preserving competitive ranking effectiveness.

cs.IR↗
arXiv2026

Distributed Seeking for Fixed Points of Biased Stochastic Operators: A Communication-Efficient Approach

This paper investigates the distributed fixed point seeking problem of sum-separable stochastic operators over the multi-agent network. Based on inexact Krasnosel'ski\uı--Mann iterations, the communication-efficient distributed algorithm is proposed under the relaxed growth bias and variance conditions, generalizing traditional unbiased and bounded additive variance assumptions. To enhance communication efficiency, we integrate communication compression and dynamic period skipping techniques, particularly adopting a unified compressor that allows both relative and absolute compression errors. By introducing a surrogate function for general non-contractive and contractive operators, we establish convergence guarantees of the distributed fixed point iteration, achieving among the first theoretical unifications with distributed non-convex optimization algorithms. Finally, numerical simulations validate the effectiveness of the theoretical results.

math.OC↗
arXiv2026

Continuous Latent Diffusion Language Model

Large language models have achieved remarkable success under the autoregressive paradigm, yet high-quality text generation need not be tied to a fixed left-to-right order. Existing alternatives still struggle to jointly achieve generation efficiency, scalable representation learning, and effective global semantic modeling. We propose Cola DLM, a hierarchical latent diffusion language model that frames text generation through hierarchical information decomposition. Cola DLM first learns a stable text-to-latent mapping with a Text VAE, then models a global semantic prior in continuous latent space with a block-causal DiT, and finally generates text through conditional decoding. From a unified Markov-path perspective, its diffusion process performs latent prior transport rather than token-level observation recovery, thereby separating global semantic organization from local textual realization. This design yields a more flexible non-autoregressive inductive bias, supports semantic compression and prior fitting in continuous space, and naturally extends to other continuous modalities. Through experiments spanning 4 research questions, 8 benchmarks, strictly matched ~2B-parameter autoregressive and LLaDA baselines, and scaling curves up to about 2000 EFLOPs, we identify an effective overall configuration of Cola DLM and verify its strong scaling behavior for text generation. Taken together, the results establish hierarchical continuous latent prior modeling as a principled alternative to strictly token-level language modeling, where generation quality and scaling behavior may better reflect model capability than likelihood, while also suggesting a concrete path toward unified modeling across discrete text and continuous modalities.

cs.CL↗
arXiv2026

Enhancing Early Detection and Localization of Gravitational Waves via Eccentricity-Induced Higher Harmonic Modes with 2G Detector Networks

Early detection and localization of gravitational waves (GWs) are essential for identifying electromagnetic (EM) counterparts, playing a key role in multi-messenger astronomy. However, second-generation (2G) ground-based detectors are most sensitive to frequencies of tens to hundreds of hertz, limiting the in-band duration of GW signals to $\mathcal{O}(0.1)$ to several tens of seconds. This constraint hinders early-warning capabilities and early localization. We present the first theoretical study on how eccentricity-induced higher harmonic modes, which enters the detector band significantly earlier than the dominant mode, enhance early detection and localization in a 2G detector network. By decomposing each harmonic mode in the frequency domain and tracking their sequential entry into the detector band, we analyze the evolution of the average signal-to-noise ratios (SNRs) and localization accuracy as functions of time-to-merger. For a GW170817-like BNS, an eccentricity of $e_0=0.4$ at 10 Hz allows the signal to reach SNR 4 and the detection threshold of SNR 8 approximately 12 and 5 minutes before merger, respectively-gains of 4.5 and 1.5 minutes over the circular case. Localization within $1000 \, (100)\,\rm deg^2$ is achievable 5 (1) minutes before merger, improving by 2 minutes (15 seconds). Our results highlight the potential of eccentricity-induced higher harmonics in improving early warnings and localization, particularly for BNS mergers, enhancing the prospects for multi-messenger astronomy.

gr-qc↗
arXiv2026

Mitigating Lost in Multi-turn Conversation via Curriculum RL with Verifiable Accuracy and Abstention Rewards

Large Language Models demonstrate strong capabilities in single-turn instruction following but suffer from Lost-in-Conversation (LiC), a degradation in performance as information is revealed progressively in multi-turn settings. Motivated by the current progress on Reinforcement Learning with Verifiable Rewards (RLVR), we propose Curriculum Reinforcement Learning with Verifiable Accuracy and Abstention Rewards (RLAAR), a framework that encourages models not only to generate correct answers, but also to judge the solvability of questions in the multi-turn conversation setting. Our approach employs a competence-gated curriculum that incrementally increases dialogue difficulty (in terms of instruction shards), stabilizing training while promoting reliability. Using multi-turn, on-policy rollouts and a mixed-reward system, RLAAR teaches models to balance problem-solving with informed abstention, reducing premature answering behaviors that cause LiC. Evaluated on LiC benchmarks, RLAAR significantly mitigates LiC performance decay (62.6% to 75.1%) and improves calibrated abstention rates (33.5% to 73.4%). Together, these results provide a practical recipe for building multi-turn reliable and trustworthy LLMs.

cs.CL↗
arXiv2026

Long-Horizon Streaming Video Generation via Hybrid Attention with Decoupled Distillation

Streaming video generation (SVG) distills a pretrained bidirectional video diffusion model into an autoregressive model equipped with sliding window attention (SWA). However, SWA inevitably loses distant history during long video generation, and its computational overhead remains a critical challenge to real-time deployment. In this work, we propose Hybrid Forcing, which jointly optimizes temporal information retention and computational efficiency through a hybrid attention design. First, we introduce lightweight linear temporal attention to preserve long-range dependencies beyond the sliding window. In particular, we maintain a compact key-value state to incrementally absorb evicted tokens, retaining temporal context with negligible memory and computational overhead. Second, we incorporate block-sparse attention into the local sliding window to reduce redundant computation within short-range modeling, reallocating computational capacity toward more critical dependencies. Finally, we introduce a decoupled distillation strategy tailored to the hybrid attention design. A few-step initial distillation is performed under dense attention, then the distillation of our proposed linear temporal and block-sparse attention is activated for streaming modeling, ensuring stable optimization. Extensive experiments on both short- and long-form video generation benchmarks demonstrate that Hybrid Forcing consistently achieves state-of-the-art performance. Notably, our model achieves real-time, unbounded 832x480 video generation at 29.5 FPS on a single NVIDIA H100 GPU without quantization or model compression. The source code and trained models are available at https://github.com/leeruibin/hybrid-forcing.

cs.CV↗
arXiv2026

Detecting gravitational wave background with equivalent configurations in the network of space based optical lattice clocks

This paper studies the use of optical lattice clock (OLC) detector networks for detecting the stochastic gravitational-wave background (SGWB). Starting from the cross-correlation formalism for two OLC detectors, we analyze how the detector geometry influences the overlap reduction function (ORF) and systematically search for configuration transformations that preserve the modulus of the ORF. We identify an equivalent transformation in which the emitting and receiving ends of both OLC links are exchanged, while the modulus of the ORF remains invariant. We then numerically compare the ORFs of isosceles trapezoidal configurations with different separations and included angles. Based on these results, we design a feasible four-spacecraft orbital configuration and evaluate its strain sensitivity and noise energy-density spectrum in comparison with LISA, Taiji, and TianQin.

gr-qc↗
arXiv2026

Standing waves for a Schrödinger system with three waves interaction

We study standing waves for a system of nonlinear Schrödinger equations with three waves interaction arising as a model for the Raman amplification in a plasma. We consider the mass-critical and mass-supercritical regimes, and we prove existence of ground states along with a synchronized mass collapse behavior. In addition, we show that the set of ground states is stable under the associated Cauchy flow. Furthermore, in the mass-supercritical setting we construct an excited state that corresponds to a strongly unstable standing wave. Moreover, a semi-trivial limiting behavior of the excited state is drawn accurately. Finally, by a refined control of the excited state's energy, we give sufficient conditions to prove global existence or blow-up of solutions to the corresponding Cauchy problem.

math.AP↗
arXiv2026

Data-driven Learning of Probabilistic Model of Binary Droplet Collision for Spray Simulation

Binary droplet collisions are ubiquitous in dense sprays. Traditional deterministic models cannot adequately represent transitional and stochastic behaviors of binary droplet collision. To bridge this gap, we developed a probabilistic model by using a machine learning approach, the Light Gradient-Boosting Machine (LightGBM). The model was trained on a comprehensive dataset of 33,540 experimental cases covering eight collision regimes across broad ranges of Weber number, Ohnesorge number, impact parameter, size ratio, and ambient pressure. The resulting machine learning classifier captures highly nonlinear regime boundaries with 99.2% accuracy and retains sensitivity in transitional regions. To facilitate its implementation in spray simulation, the model was translated into a probabilistic form, a multinomial logistic regression, which preserves 93.2% accuracy and maps continuous inter-regime transitions. A biased-dice sampling mechanism then converts these probabilities into definite yet stochastic outcomes. This work presents the first probabilistic, high-dimensional droplet collision model derived from experimental data, offering a physically consistent, comprehensive, and user-friendly solution for spray simulation.

physics.flu-dyn↗