arXiv subjects

Tong Li

Publications and source records attributed to Tong Li.

At least 37 records · Page 2Linked to original sources

arXiv2026

Continuous Spiking Graph Neural Networks

Continuous graph neural networks (CGNNs) have garnered significant attention due to their ability to generalize existing discrete graph neural networks (GNNs) by introducing continuous dynamics. They typically draw inspiration from diffusion-based methods to introduce a novel propagation scheme, which is analyzed using ordinary differential equations (ODE). However, the implementation of CGNNs requires significant computational power, making them challenging to deploy on battery-powered devices. Inspired by recent spiking neural networks (SNNs), which emulate a biological inference process and provide an energy-efficient neural architecture, we incorporate the SNNs with CGNNs in a unified framework, named Continuous Spiking Graph Neural Networks (COS-GNN). We employ SNNs for graph node representation at each time step, which are further integrated into the ODE process along with time. To enhance information preservation and mitigate information loss in SNNs, we introduce the high-order structure of COS-GNN, which utilizes the second-order ODE for spiking representation and continuous propagation. Moreover, we provide the theoretical proof that COS-GNN effectively mitigates the issues of exploding and vanishing gradients, enabling us to capture long-range dependencies between nodes. Experimental results on graph-based learning tasks demonstrate the effectiveness of the proposed COS-GNN over competitive baselines.

cs.NE↗

arXiv2026

Nonparametric Variance-Penalized Actor-Critic: Statistical Inference for Risk-Sensitive Reinforcement Learning

Variance penalization is a principled approach to risk-sensitive reinforcement learning (RL) that explicitly trades expected return for policy stability. Existing methods require a dedicated second critic to estimate return variance online, adding architectural complexity and compounding estimation error during learning. We propose a nonparametric variance-penalized actor-critic (VPAC) framework that replaces the variance critic with statistically grounded online estimators based on bootstrapping and random scaling, techniques drawn from the statistical inference literature for stochastic approximation. These estimators require no auxiliary network, maintain a single-critic architecture, and produce variance penalties that are bounded by construction, enabling clean convergence analysis. We establish almost-sure convergence for both a variance-penalized Q-learning algorithm and a two-timescale actor-critic variant via the ordinary differential equation (ODE) method, requiring only that variance estimates remain bounded rather than consistent. Empirically, we evaluate across discrete and continuous stochastic environments, demonstrating that the proposed methods match or exceed the variance reduction achieved by the existing dual-critic VPAC baseline while eliminating the overhead of a second critic. We further validate on a high-temperature superconductor (HTS) manufacturing case study, where VPAC-RS (Random Scaling) achieves a 74% reduction in steady-state critical current variability and a 63% reduction in episode return standard deviation, translating directly to improved yield consistency. Our results establish nonparametric statistical inference as a practical and theoretically sound alternative to auxiliary critics for risk-sensitive RL.

cs.LG↗

arXiv2026

"Mirror" Large Language Model Evaluations of Depression are Criterion Contaminated

Large Language Model (LLM) studies that use language responses elicited from depression assessments to predict scores on those same assessments often report near-perfect prediction of depression. We refer to these as "Mirror" evaluations and demonstrate an applied case of criterion contamination. N = 110 participants completed both structured diagnostic depression interviews (Mirror condition) and life history interviews ("Non-Mirror" condition). LLMs were prompted to predict depression scores in each condition. As expected, Mirror evaluations were near-perfect. However, Non-Mirror evaluations also displayed prediction sizes considered outstanding in psychology. Further, both Mirror and Non-Mirror predictions correlated with Patient Health Questionnaire-9 scores at similar sizes, suggesting the Mirror condition's advantage collapses when predicting an independent depression measurement. Topic modeling revealed differing depression-related themes across interview types. Mirror evaluations are better considered as reliability evaluations than as validity evaluations. Incorporating Non-Mirror approaches in LLM depression assessment may support more valid and clinically-relevant applications. Keywords: large language models, psychological assessment, psychopathology, depression, reliability, validity, criterion contamination

cs.CL↗

arXiv2026

Robust POMDP Framework for Lung Cancer Screening Problems

Lung cancer remains a leading cause of cancer mortality because many cases are diagnosed at advanced stages. Low-dose computed tomography (LDCT) screening can reduce mortality through earlier detection. Partially observable Markov decision process (POMDP) models can personalize screening by maintaining a belief over an individual's latent cancer state. However, cancer-state transition probabilities are often generated from clinical simulations and are subject to estimation error and model misspecification. We propose a robust POMDP framework using $\ell_1$-norm ambiguity sets around the nominal transition probability. The model optimizes screening decisions against the worst-case transition probability while keeping other components fixed at nominal values. Building on the piecewise-linear and convex structure of the robust value function, we adapt point-based value iteration to compute robust screening policies. We evaluate the policies using out-of-sample simulations that perturb selected cancer-progression parameters and compare them with the nominal ENGAGE policy for representative female and male heavy-smoker cohorts at age 50. Robust POMDP policies generally outperform nominal ENGAGE in mean out-of-sample quality-adjusted life-years (QALYs), with the best performance at a moderate ambiguity radius within the tested grid. Clinical analysis shows that the robust policy reduces lung cancer deaths (LCDs) in all evaluated settings for the female cohort and in most settings for the male cohort, with additional false positives (FPs). Screening-schedule analysis shows that the robust policy recommends more LDCT screens and detects more early-stage lung cancers. These findings show that incorporating transition-model uncertainty into data-driven screening models can improve out-of-sample reliability and provide more robust decision support when clinical simulation inputs are misspecified.

q-bio.QM↗

arXiv2026

Certifying Lower Bounds for Risk-Sensitive Reinforcement Learning under Adversarial State Perturbations

Reinforcement learning (RL) agents deployed in real-world environments are often vulnerable to adversarial perturbations in state observations, creating risks in safety-critical applications. Certification methods can improve robustness against adversarial perturbations by providing lower bounds on expected cumulative rewards. Existing certification methods, however, mainly focus on risk-neutral objectives. In this paper, we extend certification methods to risk-sensitive objectives by establishing lower bounds on the exponential utility of cumulative rewards under $l_{p}$-norm-bounded state adversarial perturbations ($1\leq p <\infty$). By introducing a $ϕ$-divergence relaxation of the perturbation set, we formulate the risk-sensitive certification problem as a convex optimization and derive its dual to obtain a tractable approximation of the certified lower bound. We further propose an empirical method that improves certified lower bounds by selecting the training risk-aversion parameter $β$ independently of the risk level used during evaluation. Experiments on both OpenAI Gym environments and a machine replacement problem show that, compared to risk-neutral training, risk-averse training generally yields policies with higher certified lower bounds, particularly under larger perturbation budgets. Moreover, under both risk-neutral and risk-averse evaluation settings, increasing risk aversion during training leads to non-monotonic certification performance, where certified lower bounds initially improve but eventually decrease due to overly conservative policies.

cs.LG↗

arXiv2026

Exploring millicharged particles in laboratory and astrophysical strong-field regimes

The probe of light dark particles beyond the Standard Model (SM) under a strong-field environment has drawn significant attention. In this work, we investigate the potential to search for and constrain light millicharged particles (MCPs) via strong electromagnetic fields in both laboratory laser experiments and astrophysical environments such as magnetars. We propose the MCP pair production from nonlinear Compton scattering through the interaction of a relativistic electron beam with a high-intensity laser pulse. The Furry picture and Volkov solution of Dirac equation in a background electromagnetic field are used to describe the electrons and MCPs under an external classical laser field. We calculate the cross sections of nonlinear Compton scattering to MCP pairs and take into account the irreducible SM background with missing neutrinos. We also revisit the MCP pair production via the Schwinger mechanism from magnetars with ultra-strong magnetic field and parallel electric field in polar gap. The energy loss due to the Schwinger pair production of MCPs and electric field acceleration is evaluated based on Ruderman-Sutherland model for confirmed magnetars. We find that the constraints from highly magnetized magnetars and the search potential in laboratory laser experiments are complementary.

hep-ph↗

arXiv2026

DWFF-Net: A Multi-Scale Farmland System Habitat Identification Method with Adaptive Dynamic Weight Feature Fusion

To address insufficient accuracy in multi-scale segmentation for agricultural habitat recognition, this study proposes a Dynamic Weighted Feature Fusion Network (DWFF-Net). Its encoder uses frozen DINOv3 to extract basic features and introduces a data-level adaptive dynamic weighting strategy based on relationships between image categories and feature maps. The decoder employs a dynamic weight calculation network for deep fusion of multi-level features and a hybrid loss for optimization. Statistical analysis shows that weight entropy tends to decrease as habitat category count increases, indicating adaptive adjustment of fusion strategy according to scene complexity. Experiments on a previously constructed agricultural habitat dataset validate DWFF-Net. Ablations yield mIoU 0.6979 and mF1 0.8049, exceeding the Static Weighted Feature Fusion Network by 1.82% and 1.54%, respectively, confirming that dynamic weighting improves multi-level feature utilization. Compared with U-Net, DeepLabv3+, SegFormer, and DPT, DWFF-Net improves mIoU by 16.32%, 6.49%, 4.17%, and 3.18%, respectively. For tiny features like scattered trees, IoU reaches 0.2707, outperforming those models by 99.85%, 11.45%, 22.24%, and 19.32%, verifying effectiveness in tiny habitat segmentation. This framework enables low-cost, high-precision habitat mapping and supports refined monitoring in agricultural landscapes.

cs.CV↗

arXiv2026

EComAgentBench: Benchmarking Shopping Agents on Long-Horizon Tasks with Distributed Hidden Intent

As LLM-based shopping agents enter production, existing benchmarks fail to capture how a shopper's requirements arrive: stated implicitly in the query, recorded in a profile, or revealed only when the right question is asked. Benchmarks that expose full intent upfront and grade only the final choice can neither pose this long-horizon challenge nor explain which requirement an agent missed. To address this gap, we introduce EComAgentBench, a benchmark of 662 tasks grounded in real Amazon products and reviews. Each task scatters these requirements across a visible query, a tool-gated profile, and scripted clarification; an agent must uncover hidden intent, verify candidates against attributes and review evidence, and commit to a single product within 100 tool calls. Moreover, typed, source-tagged rubrics grade every task, attributing each failure to a requirement and its source. Construction is automated yet reliable, with every answer fixed in code before any text is generated and every sample validated. Our evaluation of seven models reveals that even the strongest attains only 57.1% overall accuracy, and rubric satisfaction degrades from visible to hidden sources. Overall, we believe EComAgentBench will serve as a reproducible foundation for moving shopping agents from single-query search toward dependable assistance over long horizons.

cs.AI↗

arXiv2026

Bridging the Manifold Gap: Riemannian Residual Line Search for One-Step Image Editing

One-step diffusion editors are fast because they avoid inversion and iterative optimization, but a single transport update must be aggressive enough to realize the target prompt and conservative enough to preserve the source image--and no fixed update strength satisfies both demands across edit types. We treat this tension as a post-hoc candidate-selection problem on top of energy-field transport rather than as a new editing model. Our proposed method, Riemannian Residual Line Search, first builds a stronger edit by estimating the local time curvature of the prompt-delta field and projecting the corrected direction back onto the update norm of the original first-order energy-field transport estimation. It then forms a small residual path from the source image to this strong edit, retains the original first-order output as one candidate, and picks the final image by maximizing target-prompt CLIP alignment. On a 700-sample PIE-Bench++ evaluation across 10 edit type IDs, our method achieves state-of-the-art (SOTA) performance among current one-step update algorithms.

cs.CV↗

arXiv2026

Remember Smarter: Visual History Compressor and Hyperbolic Experience Space for Robotic Memory

Long-horizon robot policies require compact access to recent observations and reusable experience without expanding the vision-language-action (VLA) context. We introduce Remember Smarter (RS), a plug-and-play module with complementary visual-history and hyperbolic experience-memory branches. Its visual branch compresses multi-view patch histories using bidirectional spatial Mamba and causal temporal Mamba, then exposes the resulting memory to action-facing hidden states through residual cross-attention while leaving the VLM visual-token stream unchanged. Its experience branch stores successful final-layer VLM states in a Poincare VAE space, organizes them hierarchically, and asynchronously converts retrieved experience into geodesic prompt tokens without blocking action inference. When adapted to pi0, RS increases total success on LIBERO-Plus from 53.6% to 70.6% and achieves substantial performance gains in real-robot experiments designed to evaluate memory retention and experience utilization.

cs.RO↗

arXiv2026

Testing the effects of an unobservable factor: Do marriage prospects affect college major choice?

This paper develops an econometric test for whether an unobservable factor jointly influences a choice among multiple alternatives and a subsequent binary outcome. Because only transformed versions of the choice model's unobservables are identified, we combine a copula-based joint model with results on the covariance between ordered and unordered normal variables to recover the covariance between structural shocks, yielding a simulation-based test for a common unobserved factor. Applying it to the National Longitudinal Survey of Youth 1997, we find unobserved marriage prospects significantly influence college major choice. Supporting this, the dependence disappears once stated marriage expectations, elicited before major selection, are included.

econ.EM↗

arXiv2026

From Chains to Trees: Parent-Conditioned Drafting for Semi-Autoregressive Speculative Decoding

Speculative decoding accelerates LLM inference only when drafted continuations survive target-model verification. Semi-autoregressive drafters such as DSpark predict an entire token block with one backbone forward and refine it with a lightweight Markov head. However, DSpark decodes this block as a single chain, so an early mismatch invalidates the remaining suffix and limits the benefit of large draft blocks. We show that the conditional structure already learned by DSpark can support multiple parent-consistent continuations without retraining or additional backbone passes. We introduce Parent-Conditioned Drafting Tree (PCTree), which uses the pretrained Markov head to score alternative children separately for each concrete parent and allocates a fixed verification budget to the most probable paths. This converts DSpark's linear draft into a tree while preserving its one-pass parallel backbone. Across Qwen3-{4B,8B,14B} and nine benchmarks, at $B{=}7$, measured speedup gains over autoregressive (AR) decoding, relative to matched DSpark, range from $3.1\%$ to $29.5\%$. On Qwen3-4B GSM8K at $B{=}16$, PCTree increases mean acceptance length from $9.41$ to $11.16$ and three-run mean AR speedup from $6.14{\times}$ to $6.60{\times}$. These show that parent-conditioned branching can turn conditional capacity already present in a semi-autoregressive drafter into end-to-end inference gains through an inference-only change.

cs.CL↗

arXiv2026

When BBR Meets Live Streaming

Recently, industrial pioneers like Amazon, Tencent, ByteDance, and Huawei have been adopting BBR as their congestion control algorithm for live-streaming applications, including TikTok Live. However, BBR, originally crafted for bulk data transmission, faces multiple challenges in live-streaming scenarios. In this paper, we first explore two key issues associated with BBR due to inaccurate bandwidth estimation in live-streaming scenarios: (i) BBR cannot easily exit its startup phase, resulting in a fierce self-inflicted loss. (ii) BBR sends data at a lower rate than the available bandwidth during its stable phase. We then propose BBR-Copilot, an auxiliary congestion control component that cooperates with BBR, making BBR better adapt to live-streaming scenarios. BBR-Copilot allows for proactively generating accurate bandwidth measurement samples by smartly creating and sending extra data. We implement the BBR-Copilot prototype upon QUIC and evaluate it via testbed. Experimental evaluation results show that BBR-Copilot effectively enhances BBR's performance in live-streaming scenarios.

eess.IV↗

arXiv2026

Distributional Effects with Two-Sided Measurement Error: An Application to Intergenerational Income Mobility

This paper considers identification and estimation of distributional effect parameters that depend on the joint distribution of an outcome and another variable of interest ("treatment") in a setting with "two-sided" measurement error -- that is, where both variables are possibly measured with error. Examples of these parameters in the context of intergenerational income mobility include transition matrices, rank-rank correlations, and the poverty rate of children as a function of their parents' income, among others. Building on recent work on quantile regression (QR) with measurement error in the outcome (particularly, Hausman, Liu, Luo, and Palmer (2021)), we show that, given (i) two linear QR models separately for the outcome and treatment conditional on other observed covariates and (ii) assumptions about the measurement error for each variable, one can recover the joint distribution of the outcome and the treatment. Besides these conditions, our approach does not require an instrument, repeated measurements, or distributional assumptions about the measurement error. Using recent data from the 1997 National Longitudinal Study of Youth, we find that accounting for measurement error notably reduces several estimates of intergenerational mobility parameters.

econ.EM↗

arXiv2026

SF-AMS: Strategic Forgetting for Structured Memory in LLM Agent

Managing long-context dependencies remains a primary bottleneck in LLM agents, as redundant and irrelevant information can degrade multi-step reasoning. Strategic Forgetting for Agent Memory Systems (SF-AMS) is proposed as a framework for maintaining compact high-utility memory by modeling the long-term importance of memory units. SF-AMS replaces static retrieval and heuristic decay with a utility-driven survival mechanism that updates memory importance from usage redundancy and temporal signals, inducing a hierarchical memory structure that prioritizes stable entity-consistent information while filtering noise. On top of this, Composite Importance Scoring integrates semantic and entity level signals to improve retrieval robustness. Experiments on LoCoMo and LongMemEval-s show consistent gains over strong state of the art baselines including LightMem MemO and A-Mem. The largest improvement appears in multi-hop reasoning under Qwen2.5-7B where SF-AMS achieves plus 9.65 F1 over the strongest baseline followed by temporal reasoning under GPT-4o-mini plus 6.91 F1 and open-domain tasks plus 6.53 F1 demonstrating strong cross backbone generalization. These results show that modeling memory importance as a dynamic utility signal is critical for reliable long-context reasoning.

cs.AI↗

arXiv2026

The electroweak magnetic monopole in the presence of KSVZ axion

The Witten effect implies the dynamics of axion and magnetic monopole. The Cho-Maison monopole is a realistic electroweak monopole arisen in the Weinberg-Salam theory. This monopole of TeV scale mass motivates the dedicated search for electroweak monopole at colliders. In this work we investigate the implication of KSVZ axion for the electroweak magnetic monopole. We use the spherically symmetric ansatz for the electroweak dyon and introduce the spherically symmetric function for the axion field. The effective Lagrangian is then shown in terms of the electroweak monopole part, the axion kinetic energy as well as the axion interaction term. We derive the consequent equations of motion in the presence of the axion-photon coupling and show the numerical results of the topological solutions. We then calculate the changed characteristics of the electroweak monopole such as the monopole mass and the electromagnetic charges, as well as the axion potential energy.

hep-ph↗

arXiv2026

Algorithms for Adaptive Experiments that Trade-off Statistical Analysis with Reward: Combining Uniform Random Assignment and Reward Maximization

Traditional randomized A/B experiments assign arms with uniform random (UR) probability, such as 50/50 assignment to two versions of a website to discover whether one version engages users more. To more quickly and automatically use data to benefit users, multi-armed bandit algorithms such as Thompson Sampling (TS) have been advocated. While TS is interpretable and incorporates the randomization key to statistical inference, it can cause biased estimates and increase false positives and false negatives in detecting differences in arm means. We introduce a more Statistically Sensitive algorithm, TS-PostDiff (Posterior Probability of Small Difference), that mixes TS with traditional UR by using an additional adaptive step, where the probability of using UR (vs TS) is proportional to the posterior probability that the difference in arms is small. This allows an experimenter to define what counts as a small difference, below which a traditional UR experiment can obtain informative data for statistical inference at low cost, and above which using more TS to maximize user benefits is key. We evaluate TS-PostDiff against UR, TS, and two other TS variants designed to improve statistical inference. We consider results for the common two-armed experiment across a range of settings inspired by real-world applications. Our results provide insight into when and why TS-PostDiff or alternative approaches provide better tradeoffs between benefiting users (reward) and statistical inference (false positive rate and power). TS-PostDiff's adaptivity helps efficiently reduce false positives and increase statistical power when differences are small, while increasing reward more when differences are large. The work highlights important considerations for future Statistically Sensitive algorithm development that balances reward and statistical analysis in adaptive experimentation.

cs.LG↗

arXiv2026

RIE-Greedy: Regularization-Induced Exploration for Contextual Bandits

Real-world contextual bandit problems with complex reward models are often tackled with iteratively trained models, such as boosting trees. However, it is difficult to directly apply simple and effective exploration strategies--such as Thompson Sampling or UCB--on top of those black-box estimators. Existing approaches rely on sophisticated assumptions or intractable procedures that are hard to verify and implement in practice. In this work, we explore the use of an exploration-free (pure-greedy) action selection strategy, that exploits the randomness inherent in model fitting process as an intrinsic source of exploration. More specifically, we note that the stochasticity in cross-validation based regularization process can naturally induce Thompson Sampling-like exploration. We show that this regularization-induced exploration is theoretically equivalent to Thompson Sampling in the two-armed bandit case and empirically leads to reliable exploration in large-scale business environments compared to benchmark methods such as epsilon-greedy and other state-of-the-art approaches. Overall, our work reveals how regularized estimator training itself can induce effective exploration, offering both theoretical insight and practical guidance for contextual bandit design.

stat.ML↗