Searcharxiv⌕ Search

arXiv subjects

Xinyi Zhang

Publications and source records attributed to Xinyi Zhang.

At least 19 recordsLinked to original sources

EDA: Observe, Verify, and Assess for Evidence-Grounded Multimodal Depression Analysis

Existing multimodal depression assessment methods often summarize long recordings into global behavioral descriptions. This leaves the source and decision relevance of individual observations difficult to inspect. We propose EDA, an evidence-grounded framework organized around Observe, Verify, and Assess. In Observe, literature-informed acoustic and visual cues are localized into source-linked Evidence Records with temporal intervals and behavioral measurements. In Verify, frozen multimodal agents examine whether the cited media support each observation. The simplified Support admission policy retains positively verified records and their review grounds. In Assess, statistical descriptors and temporal events support reference-based behavioral comparison, while PHQ-9-guided text analysis provides complementary symptom evidence. Cross-modal disagreement triggers review of the original text before final selection. Experiments on D-Vlog and LMVD evaluate human-referenced cue agreement and downstream assessment utility separately. AI cue annotations achieve 91.02% accuracy against human consensus labels on shared source segments. With a frozen 7B backbone, EDA Support improves F1 over direct prediction by 7.61 and 9.88 percentage points on D-Vlog Leave-Query-Out and LMVD, respectively. These findings support explicit, source-linked behavioral evidence for inspectable multimodal depression assessment.

cs.AI↗

InertiaMPDD: Multimodal Personalized Depression Detection via Emotional Inertia-Inspired Dynamics Modeling

Multimodal depression screening estimates risk from speech, visual behavior and language. Current methods mainly predict labels, with limited evidence of departures from a non-depressed reference. Recording-level supervision also leaves fine-grained emotional dynamics unannotated. Motivated by emotional inertia, the persistence of affect, InertiaMPDD learns healthy next-state patterns using a world-model-inspired latent predictor. Using pretrained multimodal representations, it predicts from observed history and context. Reference deviation, residual decay rate and a Jacobian spectral descriptor form reference-derived emotional dynamics descriptors, evaluated as candidate proxy biomarkers for standalone screening and personalized decision correction. Across five cohorts from four datasets, fusion reaches 98.41% F1 on D-Vlog and 84.98% on LMVD. Standalone descriptors outperform fixed additive fusion on MPDD but discriminate weakly on AVEC2014. These findings distinguish reference informativeness from fusion benefit and reveal context-dependent utility.

cs.AI↗

Authenticated Approximate Nearest Neighbor Search in Outsourced Vector Databases

Approximate Nearest Neighbor (ANN) search has become a fundamental query primitive in modern vector databases for managing large-scale embedding data. To support growing vector collections, organizations increasingly outsource database management and query processing to third-party clouds. However, untrusted service providers may return suboptimal or fabricated results, posing a serious threat to ANN search integrity. Existing authenticated ANN approaches face substantial scalability challenges. They typically require transmitting high-dimensional vectors for distance verification and rely on topology-agnostic authentication structures, resulting in excessive communication and verification overhead. In this paper, we propose VIDA, a novel and efficient scheme for authenticated graph-based ANN search. VIDA replaces raw vector transmission with compact vector commitments and employs an aggregated Inner Product Argument to efficiently verify multiple distance computations. Furthermore, we introduce MeST, a topology-aware authenticated data structure that aligns authentication paths with graph traversal, significantly reducing structural proof overhead. Additionally, we design a quantization-aware authentication strategy that exploits distance bounds for pruning, reducing the number of exact distances requiring cryptographic authentication. Extensive experiments demonstrate that the proposed schemes substantially improve the efficiency of authenticated ANN search, achieving up to a $6.6\times$ speedup in end-to-end latency and reducing data transfer time by up to 99.2% compared with state-of-the-art methods.

cs.DB↗

VoxLution: Evolving Voxels for High-Fidelity 3D Scene Generation

High-resolution 3D indoor scene generation is crucial for immersive applications such as virtual reality, spatial computing, and robotics simulation. We introduce VoxLution, a coarse-to-fine transformer-based approach that generates high-fidelity indoor scenes with rich details, overcoming the overly smooth geometry typical of existing generative models. First, we generate a low-resolution spatial layout; then, based on this structural information, we progressively refine the scene geometry locally in chunks. Key to our approach is formulating 3D super-resolution as a sequence modeling problem over multi-resolution sparse voxel grids. To achieve this, we introduce a training-free, resolution-invariant tokenization scheme that compactly encodes the geometric information of eight octants into a single decimal number. We formulate the transformer sequence to include multiple resolutions, allowing the coarse layout to guide the generation of fine local details. To ensure coherence across local chunks, we incorporate adjacent chunk boundary information directly into the transformer sequencing. VoxLution achieves scene generation at a finer voxel resolution than existing baselines, demonstrating state-of-the-art performance on standard benchmarks with 59.6% lower FPD and 51.1% lower FID on 3D-FRONT, and 70.5% lower FPD and 73.8% lower FID on Scannet++.

cs.CV↗

Probing Red Giant Interiors with g-Dominated Mixed Modes II: The Cases of KIC 6117517, KIC 9475697, KIC 10123207 and KIC 10200377

Building on Paper~I, which constrained core overshooting to $f_{\rm ov} < 0.01$ for $M < 1.4\,M_{\odot}$ and found $α_{\rm MLT} \simeq 2.0$ for metal-rich red giants, we extend this analysis to four lower-mass \textit{Kepler} red giants spanning a broad metallicity range: KIC~6117517, KIC~9475697, KIC~10123207, and KIC~10200377. Two complementary model grids are constructed: Grid~A fixes $α_{\rm MLT} = 2.0$ while varying metallicity; Grid~B adopts spectroscopic metallicities with $α_{\rm MLT}$ as a free parameter. Grid~A successfully reproduces both asteroseismic and spectroscopic constraints for the metal-rich targets (KIC~6117517 and KIC~9475697), confirming $α_{\rm MLT} = 2.0$ for metal-rich red giants. For the metal-poor targets (KIC~10123207 and KIC~10200377), Grid~B yields larger mixing-length parameters of $α_{\rm MLT} \approx 2.16$--$2.18$. Although only four stars are analyzed with unequal sensitivity to $α_{\rm MLT}$, these results suggest a tentative increase in convective efficiency toward lower metallicity, consistent with 3D hydrodynamical simulations but opposite to some 1D empirical calibrations. The best-fitting overshooting parameters ($f_{\rm ov} = 0.006$--$0.007$) independently support the Paper~I constraint and are consistent with a mass-dependent overshooting relation. The parameter uncertainties reach $0.6$--$2.5\%$ in mass, $0.2$--$0.9\%$ in radius, $2$--$13\%$ in age, and below $0.6\%$ in helium-core mass and below $0.2\%$ in helium-core radius. Surface-effect corrections are negligible for g-dominated mixed modes, validating $χ^2_{l=1g}$ as a robust fitting metric. For KIC~9475697, the core rotation rate is determined to be $Ω_{\rm core} = 0.6945 \pm 0.0190\,μ$Hz from model-based trapping parameters.

astro-ph.SR↗

More efficient transportation of causal effects using mediators

We develop flexible semiparametric estimators of the average treatment effect transported to a target population that leverage mediators to achieve identification, and prior knowledge of population heterogeneity or summary-level statistics to achieve improved efficiency. We first propose a one-step semiparametric estimator that assumes knowledge of the baseline covariates driving the heterogeneity across populations. This approach improves efficiency by transporting only through these covariates, rather than through the full covariate set. We then introduce a collaborative one-step semiparametric estimator that also improves efficiency without requiring knowledge of which covariates drive study heterogeneity. Collectively, these estimators advance our ability to transport causal effects by improving efficiency in the presence of mediation and intermediate confounding through targeted dimension reduction strategies. We use simulation to examine finite-sample performance and apply our estimators to a large-scale housing mobility trial (MTO) and a hearing intervention trial (ACHIEVE).

stat.ME↗

Convergence of Half-Space Last Passage Percolation Away from the Boundary to the Directed Landscape

In this note, we prove convergence of the half-space exponential last passage percolation (LPP) model, away from the boundary, to the directed landscape. Our approach couples the half-space and full-space LPP models and constructs two barrier events based on the monotonicity of last passage paths. Combining this coupling with moderate deviation estimates for both models and the known convergence of full-space LPP to the directed landscape, we establish the desired convergence.

math.PR↗

Bulk Directed Landscape Convergence and Phase Transitions in Half-Space Models

We study the effects of the boundary in the bulk of two half-space models. For half-space exponential last passage percolation (LPP) and the half-space log-gamma polymer, we establish convergence of their respective rescaled last passage time and free energy to the directed landscape when the observation window is at distance $f(n)n^{2/3}$ from the boundary and the attractive boundary strength is $o(f(n)n^{-1/3})$, where $f(n)\to\infty$ may diverge arbitrarily slowly. We also show that this condition on the boundary strength is sharp at the level of scales: any boundary strength of order $f(n)n^{-1/3}$ changes the limiting behavior for sufficiently large temporal separations. For exponential LPP, we prove a phase transition in the point-to-point last passage time as the boundary attraction varies on the scale \(f(n)n^{-1/3}\). As the attraction increases, the fluctuations transition from a bulk-dominated KPZ regime to a boundary-dominated Gaussian regime, with a critical window governed by the positive part of a shifted Gaussian. The limiting law in the critical window manifests the competition between a Gaussian boundary contribution and a bulk contribution whose magnitude vanishes under the Gaussian scaling.

math.PR↗

Gauge Interactions Involving New Vector and Pseudoscalar Particles

The Vector gauge bosons $Z'$ and axion-like particles (ALPs) $a$ are predicted by the $U(1)_X$ gauge extension and Peccei-Quinn (PQ) models, respectively. We focus on the interactions of the $Z'$ boson and the $a$ particle with the $W^\pm$ gauge boson, and investigate the production cross section and related kinematic distributions for the $Z'/a\to W^+W^-$ decay process at hadron colliders. We consider the angular distributions of charged leptons in the final states corresponding to the semileptonic and fully leptonic decays of the $W^\pm$ boson, separately. The study shows that, in the associated production processes of $Z'/a$ with $W^+$ or $Z$ bosons, the angular distributions of final-state leptons can serve as an important observable for probing new gauge interactions, and can also provide a criterion to distinguish vector particles from pseudoscalar particles.

hep-ph↗

World-T3: Test-Time Training as Persistent Memory for Real-Time Video World Models

Recent advances in video world models have enabled plausible interactive environments, yet long-horizon generation places a fundamental demand on visual memory: the previously observed world must remain consistent as the model continues to generate. Existing autoregressive approaches often implicitly rely on the generative model itself to retain such information, leading to spatial memory decay beyond the active context. We introduce World-T3, a camera-controlled video world model with a test-time training memory that consolidates past observations into dedicated model parameters. To robustly update this memory, we propose a recursive least-squares rule that incrementally absorbs novel visual features via online pose-keyed memory updates. Our TTT memory reads out previously observed scene features from the fast-weight state by camera pose, enabling consistent generation with constant storage and per-step computation regardless of rollout length. To facilitate real-time interactive generation with this robust memory, we further distill the autoregressive model into a few-step generator via dual-teacher distribution matching. On a long-horizon revisit benchmark, World-T3 raises revisit PSNR from 13.52 to 18.31 dB and reduces absolute rotation error from 3.344 to 1.075 relative to the strongest baseline on both metrics, while sustaining 20.6 FPS at 1280 x 704 on a single NVIDIA B200 GPU.

cs.CV↗

The discrete periodic Pitman transform: invariances, braid relations, and Burke properties

We develop the theory of the discrete periodic Pitman transform, first introduced by Corwin, Gu, and the fifth author. We prove that the discrete periodic Pitman transform satisfies the same braid relations that are satisfied for the full-line Pitman transform shown by Biane, Bougerol, and O'Connell. This defines a group action of the infinite symmetric group on sequences of vectors in $\mathbb R^{\mathbb Z_N}$. We prove that, for polymers in a periodic environment, single-path and multi-path partition functions are preserved under the action of this transform on the weights in the polymer model. Combined with a new inhomogeneous Burke property for the periodic Pitman transform, we prove a multi-path invariance result for the periodic inverse-gamma polymer under permutations of the column parameters. In the limit to the full-line case, we obtain a multi-path extension of a recent invariance result of Bates, Emrah, Martin, Seppäläinen, and the fifth author, in both positive and zero-temperature.

math.PR↗

Train Ahead, Distill Back: Bootstrapping On-Policy Self-Distillation for Large Language Models

On-policy self-distillation (OPSD) improves large language models by letting a self-teacher with privileged information provide dense token-level supervision on the model's own trajectories. Yet existing methods typically construct the self-teacher from the current, initial, or slowly averaged policy state, leaving the quality of supervision constrained by the teacher's ability to exploit privileged information. We ask whether the model's own optimization progress can instead be recycled into a stronger self-teacher. In this paper, we introduce Bootstrapped On-Policy Self-Distillation (B-OPSD), which temporarily trains the policy ahead to obtain a future teacher, restores the student to the original policy state, and then uses the future teacher to supervise the restarted student. The future teacher improves supervision in two complementary ways, it can generate more reliable privileged trajectories and, conditioned on them, provide more informative token-level targets along the restarted student's on-policy trajectories. Experiments on mathematical reasoning with Qwen3-4B and Qwen3-8B show consistent improvements over standard OPSD in both settings, including gains from 27.50 to 41.30 and from 48.80 to 64.44 in the rollout-privileged setting. Our findings point to a broader principle for self-improving models that future learning progress can be distilled backward, preserving acquired knowledge while bootstrapping beyond the optimization state that produced it.

cs.AI↗

TensorHub: Scalable and Elastic Weight Transfer for LLM RL Training

Modern LLM reinforcement learning (RL) workloads require a high-performance weight transfer system to scale training across heterogeneous compute resources. However, efficiently transferring terabyte-scale model weights across thousands of GPUs remains challenging because the system must accommodate clusters that dynamically scale up and down while keeping coordination, data movement, and storage overhead low. We introduce Reference-Oriented Storage (ROS), a new storage abstraction for RL weight transfer that exploits highly replicated model weights in place. ROS presents the illusion that certain versions of the model weights are stored and can be fetched on demand. Underneath, ROS does not physically store any copies of the weights; instead, it tracks the workers that hold these weights on GPUs for inference. Upon request, ROS directly uses them to serve reads. We build TensorHub, a production-quality system that instantiates the ROS idea with topology-aware transfer, model-parallel consistency, and fault tolerance. Evaluation shows that TensorHub saturates RDMA bandwidth and adapts to three distinct rollout workloads with minimal engineering effort. Specifically, TensorHub reduces total GPU stall time by up to 6.7x for standalone rollouts, accelerates weight updates for elastic rollouts by up to 4.8x, and cuts cross-datacenter rollout stall time by up to 19x. TensorHub has been deployed in ByteDance production to support cutting-edge RL training.

cs.DC↗

Deep Reinforcement Learning for Equity Trading: Benchmarking Actor-Critic Methods with Forward Retraining

Consistently profitable trading is difficult because equity markets are noisy, non-stationary, and only partially predictable from historical data. We benchmark five deep reinforcement learning (DRL) actor-critic methods: A2C, PPO, DDPG, TD3, and SAC, that learn trading actions end-to-end from market states, and compare them with a supervised price-forecasting baseline. Using daily data for 20 large-capitalization S&P 500 stocks from 2000 to 2020, enriched with trend-following technical indicators and log min-max scaling, we train on 2000-2018 and backtest on 2019-2020. Each agent is evaluated both when trained once and under forward retraining, in which it is retrained on all data available before each successive test window. DDPG achieves the highest annual return (55.5%), Sharpe ratio (1.38), and alpha (0.22), but also the highest market beta (1.24). TD3 and SAC offer a better risk-return balance, with Sharpe ratios of 1.37 and 1.33 and maximum drawdowns of about 25%. Forward retraining improves A2C, PPO, and SAC, leaves TD3 essentially unchanged, and reduces DDPG's annual return from 55.5% to 29.8%, consistent with TD3's greater robustness to hyperparameters. The forecasting baseline has the smallest maximum drawdown (9.6%) and the lowest beta (0.31), underscoring a trade-off between the higher returns of end-to-end DRL and the lower risk of forecast-driven strategies.

cs.LG↗

Leptogenesis from the Dirac CP-violating phase in the minimal left-right symmetric model

Leptogenesis from low-energy CP violation provides a vital link between neutrino physics and the observed baryon asymmetry of the universe. However, this connection is typically obscured by unknown high-energy parameters. In this work, we investigate thermal leptogenesis in the Minimal Left-Right Symmetric Model with generalized parity as the left-right symmetry, where the hermiticity of the Dirac neutrino coupling allows the right-handed mixing matrix $V_\mathrm{R}$ to be determined with minimal assumptions. We show that for a real $V_\mathrm{R}$, these conditions favor CP-conserving Majorana phases, leaving the Dirac CP-violating phase ($δ$) as the sole source of asymmetry. By numerically exploring all four leptogenesis scenarios, we demonstrate that $δ$ alone can generate the observed baryon asymmetry with the correct sign within specific regions of the parameter space. The results exhibit a high sensitivity to the neutrino mass ordering and the lightest neutrino mass, providing a stringent, testable framework for future experimental measurements of the CP phase and neutrino mass scale.

hep-ph↗

SG-UMP: Sequence-Guided Universal Multimodal Prioritization Calculation Framework

Multimodal sequential recommendation (MSR) improves recommendation by incorporating heterogeneous information such as text, images, and user interactions. However, existing MSR methods often fail to capture user-level preference heterogeneity and dataset-level modality bias, limiting their adaptability across users and datasets. To address this issue, we propose \textbf{S}equence-\textbf{G}uided \textbf{U}niversal \textbf{M}ultimodal \textbf{P}rioritization Calculation Framework (\textbf{SG-UMP}), a plug-and-play plugin for enhancing multimodal information processing in MSR. SG-UMP includes a Module Combiner for flexible multimodal processing and a Module Router for dynamic module ordering, enabling adaptation to both user preferences and dataset characteristics. Experiments on four real-world datasets show that SG-UMP consistently improves recommendation performance across different backbones and multimodal settings. The code is available at https://github.com/esemsc-xz524/SG-UMP .

cs.IR↗

StartupBench: Benchmarking General-Purpose Agents on Market-Validated End-to-End Workflows

Recent advances in Large Language Models(LLMs) and agents have substantially improved the ability of AI systems to execute complex tasks. Yet existing benchmarks largely rely on researcher-selected tasks, leaving uncertain whether such progress extends to the work that real-world users actually demand from AI systems. We introduce \textbf{StartupBench}, an E2E agent benchmark grounded in market-validated AI startup products. Rather than defining tasks from pre-defined assumptions about useful agent capabilities, we systematically study AI products with demonstrated adoption, together with their product workflows and users, to identify real-world tasks for which AI has established practical demand across diverse professional domains. We translate these workflows into complete deliverable-oriented tasks and evaluate them with fine-grained rubrics capturing their complex requirements. Across representative models evaluated under a unified agent harness, even the strongest model successfully completes only approximately 30\% of StartupBench, despite making substantial partial progress on many tasks. Further analysis identifies aspects like complex instruction following and domain-specific expertise as major sources of failure. Our results reveal that many market-validated workflows remain beyond the reliable capabilities of current general-purpose agents, establishing StartupBench as an empirical measure of progress toward E2E completions of real-world user tasks.

cs.AI↗

GAINS: Leveraging Inconsistent Human Intervention Signals in Reinforcement Learning

Correcting robot manipulation policies through human intervention holds great promise for real-world deployment, yet human operators are inherently imperfect in both the actions they provide and the timing of their intervention signals. While the former has been extensively discussed in reinforcement learning (RL), the latter remains underexplored. At high control frequencies, human intervention signals are often delayed and inconsistent across time and state space. In this work, we present GAINS, a framework for leveraging inconsistent human intervention signals in RL. At the core of GAINS, we employ distributional RL with quantile Q-networks to model the return variability induced by sparse task rewards and inconsistent human interventions. Building on this distributional representation, we introduce a pessimistic exploration strategy that promotes safe and sample-efficient learning under human corrections. We evaluate GAINS on four diverse simulated manipulation tasks and two challenging real-world scenarios against state-of-the-art intervention-based methods. GAINS achieves a 22% higher task success rate than RLIF and improves recovery success by up to 43% in failure scenarios. These results highlight the importance of modeling return variability induced by human imperfection for real-world deployment of intervention-based learning.

cs.RO↗