arXiv subjects

Yi Li

Publications and source records attributed to Yi Li.

At least 127 records · Page 7Linked to original sources

arXiv2026

The Calabi flow with prescribed curvature on finite graphs

In this paper, we investigate the prescribed curvature problem associated with a special Lin-Lu-Yau curvature on finite graphs of girth at least 6. We define the corresponding Calabi flow for this curvature type, and establish an equivalent characterization of the problem, namely, the solution to the Calabi flow exists globally in time and converges if and only if there exists a weight function that realizes the prescribed curvature. In particular, for constant curvature weights, we prove that the solution to the Calabi flow exists globally in time and converges under certain topological conditions.

math.DG↗

arXiv2026

EgoReasoner: Learning Egocentric 4D Reasoning via Task-Adaptive Structured Thinking

Egocentric video understanding is inherently complex due to the dynamic 4D nature of the environment, where camera motion and object displacements necessitate a continuous re-evaluation of spatial relations. In this work, we target a suite of under-explored egocentric 4D reasoning tasks, including fixture interaction counting, viewpoint-relative fixture location, object movement itinerary tracking, and stationary object localization, that require fundamentally different cognitive operations: spatial anchoring, temporal tracking, and duration reasoning. We observe that these structural differences make task-agnostic approaches insufficient: generic Chain-of-Thought methods lack task-appropriate reasoning primitives, and uniform reinforcement learning actively destabilizes performance on spatial tasks. To address this, we propose EgoReasoner, a two-stage framework that aligns both the reasoning scaffold and the reward signal to each task's cognitive structure. In the first stage, Task-Adaptive Thinking Templates guide the synthesis of structured CoT traces that teach the model to reason adaptively across task types via supervised fine-tuning. In the second stage, task-aware reward functions verify entity grounding, temporal alignment, and task-adaptive logical consistency, selectively strengthening each reasoning pathway via reinforcement fine-tuning with GRPO. Our 3B-parameter model, trained on only 16K samples, achieves 37.5% average accuracy on the challenging HD-EPIC benchmark, surpassing Qwen2.5-VL-7B (25.7%) by over 10 points.

cs.CV↗

arXiv2026

Beyond Consistency: Inference for the Relative risk functional in Deep Nonparametric Cox Models

There remain theoretical gaps in deep neural network estimators for the nonparametric Cox proportional hazards model. In particular, it is unclear how gradient-based optimization error propagates to population risk under partial likelihood, how pointwise bias can be controlled to permit valid inference, and how ensemble-based uncertainty quantification behaves under realistic variance decay regimes. We develop an asymptotic distribution theory for deep Cox estimators that addresses these issues. First, we establish nonasymptotic oracle inequalities for general trained networks that link in-sample optimization error to population risk without requiring the exact empirical risk optimizer. We then construct a structured neural parameterization that achieves infinity-norm approximation rates compatible with the oracle bound, yielding control of the pointwise bias. Under these conditions and using the Hajek--Hoeffding projection, we prove pointwise and multivariate asymptotic normality for subsampled ensemble estimators. We derive a range of subsample sizes that balances bias correction with the requirement that the Hajek--Hoeffding projection remain dominant. This range accommodates decay conditions on the single-overlap covariance, which measures how strongly a single shared observation influences the estimator, and is weaker than those imposed in the subsampling literature. An infinitesimal jackknife representation provides analytic covariance estimation and valid Wald-type inference for relative risk contrasts such as log-hazard ratios. Finally, we illustrate the finite-sample implications of the theory through simulations and a real data application.

stat.ML↗

arXiv2026

CAMO: A Conditional Neural Solver for the Multi-objective Multiple Traveling Salesman Problem

Robotic systems often require a team of robots to collectively visit multiple targets while optimizing competing objectives, such as total travel cost and makespan. This setting can be formulated as the Multi-Objective Multiple Traveling Salesman Problem (MOMTSP). Although learning-based methods have shown strong performance on the single-agent TSP and multi-objective TSP variants, they rarely address the combined challenges of multi-agent coordination and multi-objective trade-offs, which introduce dual sources of complexity. To bridge this gap, we propose CAMO, a conditional neural solver for MOMTSP that generalizes across varying numbers of targets, agents, and preference vectors, and yields high-quality approximations to the Pareto front (PF). Specifically, CAMO consists of a conditional encoder to fuse preferences into instance representations, enabling explicit control over multi-objective trade-offs, and a collaborative decoder that coordinates all agents by alternating agent selection and node selection to construct multi-agent tours autoregressively. To further improve generalization, we train CAMO with a REINFORCE-based objective over a mixed distribution of problem sizes. Extensive experiments show that CAMO outperforms both neural and conventional heuristics, achieving a closer approximation of PFs. In addition, ablation results validate the contributions of CAMO's key components, and real-world tests on a mobile robot platform demonstrate its practical applicability.

cs.RO↗

arXiv2026

Inference for Deep Neural Network Estimators in Generalized Nonparametric Models

While deep neural networks (DNNs) are used for prediction, inference on DNN-estimated subject-specific means for categorical or exponential family outcomes remains underexplored. We address this by proposing a DNN estimator under generalized nonparametric regression models (GNRMs) and developing a rigorous inference framework. Unlike existing approaches that assume independence between estimation errors and inputs to establish the error bound, a condition often violated in GNRMs, we allow for dependence and our theoretical analysis demonstrates the feasibility of drawing inference under GNRMs. To implement inference, we consider an Ensemble Subsampling Method (ESM) that leverages U-statistics and the Hoeffding decomposition to construct reliable confidence intervals for DNN estimates. We show that, under GNRM settings, ESM enables model-free variance estimation and accounts for heterogeneity among individuals in the population. Through simulations under nonparametric logistic, Poisson, and binomial regression models, we demonstrate the effectiveness and efficiency of our method. We further apply the method to the electronic Intensive Care Unit (eICU) dataset, a large scale collection of anonymized health records from ICU patients, to predict ICU readmission risk and offer patient-centric insights for clinical decision making.

stat.ML↗

arXiv2026

Switched Linear Ensemble Systems and Structural Controllability

This paper introduces and solves a structural controllability problem for ensembles of switched linear systems. All individual systems in the ensemble are sparse and governed by the same sparsity pattern, and undergo switching among subsystems by following the same switching sequence. The controllability of an ensemble system describes the ability to use a common control input to simultaneously steer every individual system. A sparsity pattern is called structurally controllable for pair $(k,q)$ if it admits a controllable ensemble of $q$ individual systems with at most $k$ subsystems. We derive a necessary and sufficient condition for a sparsity pattern to be structurally controllable for a given $(k,q)$, and characterize when a sparsity pattern admits a finite $k$ that guarantees structural controllability for $(k,q)$ for arbitrary $q$. Compared with the linear time-invariant ensemble case, this second condition is strictly weaker. We further show that these conditions have natural connections with maximum flow, and hence can be checked by polynomial algorithms. Specifically, the time complexity of deciding structural controllability is $O(n^3)$ and the complexity of computing the smallest number of subsystems needed is $O(n^3 \log n)$, with $n$ the dimension of each individual system.

math.OC↗

arXiv2026

When Semantics Connect the Swarm: LLM-Driven Fuzzy Control for Cooperative Multi-Robot Underwater Coverage

Underwater multi-robot cooperative coverage remains challenging due to partial observability, limited communication, environmental uncertainty, and the lack of access to global localization. To address these issues, this paper presents a semantics-guided fuzzy control framework that couples Large Language Models (LLMs) with interpretable control and lightweight coordination. Raw multimodal observations are compressed by the LLM into compact, human-interpretable semantic tokens that summarize obstacles, unexplored regions, and Objects Of Interest (OOIs) under uncertain perception. A fuzzy inference system with pre-defined membership functions then maps these tokens into smooth and stable steering and gait commands, enabling reliable navigation without relying on global positioning. Then, we further coordinate multiple robots by introducing semantic communication that shares intent and local context in linguistic form, enabling agreement on who explores where while avoiding redundant revisits. Extensive simulations in unknown reef-like environments show that, under limited sensing and communication, the proposed framework achieves robust OOI-oriented navigation and cooperative coverage with improved efficiency and adaptability, narrowing the gap between semantic cognition and distributed underwater control in GPS-denied, map-free conditions.

cs.RO↗

arXiv2026

OneRanker: Unified Generation and Ranking with One Model in Industrial Advertising Recommendation

The end-to-end generative paradigm is revolutionizing advertising recommendation systems, driving a shift from traditional cascaded architectures towards unified modeling. However, practical deployment faces three core challenges: the misalignment between interest objectives and business value, the target-agnostic limitation of generative processes, and the disconnection between generation and ranking stages. Existing solutions often fall into a dilemma where single-stage fusion induces optimization tension, while stage decoupling causes irreversible information loss. To address this, we propose OneRanker, achieving architectural-level deep integration of generation and ranking. First, we design a value-aware multi-task decoupling architecture. By leveraging task token sequences and causal mask, we separate interest coverage and value optimization spaces within shared representations, effectively alleviating target conflicts. Second, we construct a coarse-to-fine collaborative target awareness mechanism, utilizing Fake Item Tokens for implicit awareness during generation and a ranking decoder for explicit value alignment at the candidate level. Finally, we propose input-output dual-side consistency guarantees. Through Key/Value pass-through mechanisms and Distribution Consistency (DC) Constraint Loss, we achieve end-to-end collaborative optimization between generation and ranking. The full deployment on Tencent's WeiXin channels advertising system has shown a significant improvement in key business metrics (GMV - Normal +1.34\%), providing a new paradigm with industrial feasibility for generative advertising recommendations.

cs.IR↗

arXiv2026

Accelerating Scientific Research with Gemini: Case Studies and Common Techniques

Recent advances in large language models (LLMs) have opened new avenues for accelerating scientific research. While models are increasingly capable of assisting with routine tasks, their ability to contribute to novel, expert-level mathematical discovery is less understood. We present a collection of case studies demonstrating how researchers have successfully collaborated with advanced AI models, specifically Google's Gemini-based models (in particular Gemini Deep Think and its advanced variants), to solve open problems, refute conjectures, and generate new proofs across diverse areas in theoretical computer science, as well as other areas such as economics, optimization, and physics. Based on these experiences, we extract common techniques for effective human-AI collaboration in theoretical research, such as iterative refinement, problem decomposition, and cross-disciplinary knowledge transfer. While the majority of our results stem from this interactive, conversational methodology, we also highlight specific instances that push beyond standard chat interfaces. These include deploying the model as a rigorous adversarial reviewer to detect subtle flaws in existing proofs, and embedding it within a "neuro-symbolic" loop that autonomously writes and executes code to verify complex derivations. Together, these examples highlight the potential of AI not just as a tool for automation, but as a versatile, genuine partner in the creative process of scientific discovery.

cs.CL↗

arXiv2026

On Pseudo-Effectivity and Volumes of Adjoint Classes in Kähler Families with Projective Central Fiber

This paper is devoted to studying the deformation behavior of pseudo-effective canonical divisors and volumes of adjoint classes in Kähler families. Based on recent developments in the Kähler minimal model program, for flat families with fiberwise canonical singularities, we establish the global stability of the pseudo-effectivity of canonical divisors and uniruledness, assuming in addition that one fiber is projective, while the same conclusion for Kähler threefolds is also true without the projectivity assumption of the central fiber. For a smooth Kähler family whose central fiber is projective with a big adjoint class, we show that its volume remains locally constant. Finally, using the (relative) minimal model program for Kähler threefolds, we verify the deformation invariance of volumes of adjoint classes and plurigenera for smooth families of Kähler threefolds, thereby confirming Siu's invariance of plurigenera conjecture in dimension three.

math.AG↗

arXiv2026

Fed-GAME: Personalized Federated Learning with Graph Attention Mixture-of-Experts For Time-Series Forecasting

Federated learning (FL) on graphs shows promise for distributed time-series forecasting. Yet, existing methods rely on static topologies and struggle with client heterogeneity. We propose Fed-GAME, a framework that models personalized aggregation as message passing over a learnable dynamic implicit graph. The core is a decoupled parameter difference-based update protocol, where clients transmit parameter differences between their fine-tuned private model and a shared global model. On the server, these differences are decomposed into two streams: (1) averaged difference used to updating the global model for consensus (2) the selective difference fed into a novel Graph Attention Mixture-of-Experts (GAME) aggregator for fine-grained personalization. In this aggregator, shared experts provide scoring signals while personalized gates adaptively weight selective updates to support personalized aggregation. Experiments on two real-world electric vehicle charging datasets demonstrate that Fed-GAME outperforms state-of-the-art personalized FL baselines.

cs.LG↗

arXiv2026

Modulating Surface Acoustic Wave Generation through Superconductivity

Surface acoustic waves (SAWs), with their five orders-of-magnitude slower propagation velocity, allow for considerably shorter wavelengths at the same frequency compared to electromagnetic waves. The short wavelengths allow for device miniaturization and on-chip integration. The generic design of these devices involve piezoelectric substrates with comblike arrays of Al or Au electrodes known as interdigitated transducers deposited on the surface. However, Al and Au both have shortcomings at the cryogenic temperatures required for quantum applications, namely the formation of two-level systems and the lack of superconductivity perpetuating Ohmic losses, respectively. In this work, SAWs are generated in the high-MHz to low-GHz range using niobium nitride (NbN) interdigitated transducers (IDTs) and Bragg reflectors. We demonstrate the fabrication of acoustic devices through photolithography and reactive ion etching (RIE). The sharp transition between superconducting and normal states and the corresponding change in SAW transmission allows for fine control of the 'on' (superconducting) and 'off' (normal) states of NbN, with a Δ_T = K separating the transmission minimum and maximum. We demonstrate a 16x difference in transmission between the 'on' and 'off' states of the device. The SAW transmission behavior mirrors the change in resistance of NbN at its Tc. These findings open up new possibilities for the integration of NbN SAW resonators into existing quantum architectures based on NbN and a method for adjusting transmission properties independent of applied voltage.

cond-mat.mes-hall↗

arXiv2026

A Physics-Guided Neural Framework for Rheology Measurement from Dynamical Laser Speckles

Critical breakthroughs in the area of biomedicine and materials science increasingly depend on rapid, non-contact methods for viscoelastic characterization. Laser Speckle Rheology (LSR) is positioned to meet this demand, effectively circumventing the speed and invasiveness bottlenecks inherent to traditional mechanical rheometer. However, its application in turbid fluids is severely constrained by multiple scattering, where standard physical inversions rely heavily on precise, sample-specific optical transport parameters that are difficult to measure in situ. To overcome this barrier, we propose a physics-guided deep learning framework that infers a Maxwell relaxation spectrum from the intensity autocorrelation g2(t) and speckle-intensity histogram statistics. The resulting spectrum is then propagated through a Maxwell forward model to predict G'and G'' under physics-consistency constraints. Quantitatively, the framework achieves RMSElog as low as 0.009 against reference and generalizes to previously unseen scattering conditions, preserving physically plausible frequency dependence and G'- G'' phase behavior. It reduces reliance on optical transport parameters that are hard to determine in situ and returns an interpretable generalized Maxwell relaxation spectrum, improving the practicality of LSR in turbid media.

physics.optics↗

arXiv2026

HAL-MLE Log-Splines Density Estimation (Part I: Univariate)

We study nonparametric maximum likelihood estimation of probability densities under a total variation (TV) type penalty, sectional variation norm (also named as Hardy-Krause variation). TV regularization has a long history in regression and density estimation, including results on $L^2$ and KL divergence convergence rates. Here, we revisit this task using the Highly Adaptive Lasso (HAL) framework. We formulate a HAL-based maximum likelihood estimator (HAL-MLE) using the log-spline link function from \citet{kooperberg1992logspline}, and show that in the univariate setting the bounded sectional variation norm assumption underlying HAL coincides with the classical bounded TV assumption. This equivalence directly connects HAL-MLE to existing TV-penalized approaches such as local adaptive splines \citep{mammen1997locally}. We establish three new theoretical results: (i) the univariate HAL-MLE is asymptotically linear, (ii) it admits pointwise asymptotic normality, and (iii) it achieves uniform convergence at rate $n^{-(k+1)/(2k+3)}$ up to logarithmic factors for the smoothness order $k \geq 1$. These results extend existing results from \citet{van2017uniform}, which previously guaranteed only uniform consistency without rates when $k=0$. We will include the uniform convergence for general dimension $d$ in the follow-up work of this paper. The intention of this paper is to provide a unified framework for the TV-penalized density estimation methods, and to connect the HAL-MLE to the existing TV-penalized methods in the univariate case, despite that the general HAL-MLE is defined for multivariate cases.

math.ST↗

arXiv2026

Hippocampus: An Efficient and Scalable Memory Module for Agentic AI

Agentic AI require persistent memory to store user-specific histories beyond the limited context window of LLMs. Existing memory systems use dense vector databases or knowledge-graph traversal (or hybrid), incurring high retrieval latency and poor storage scalability. We introduce Hippocampus, an agentic memory management system that uses compact binary signatures for semantic search and lossless token-ID streams for exact content reconstruction. Its core is a Dynamic Wavelet Matrix (DWM) that compresses and co-indexes both streams to support ultra-fast search in the compressed domain, thus avoiding costly dense-vector or graph computations. This design scales linearly with memory size, making it suitable for long-horizon agentic deployments. Empirically, our evaluation shows that Hippocampus reduces end-to-end retrieval latency by up to 31$\times$ and cuts per-query token footprint by up to 14$\times$, while maintaining accuracy on both LoCoMo and LongMemEval benchmarks.

cs.AI↗

arXiv2026

Magnonic spontaneous oscillation induced by parametric pumping

Spontaneous dynamic systems have attracted significant attention for their rich underlying physics such as phase-locking and synchronization. In this work, we report a new mechanism of generating magnetic spontaneous oscillation via parametric pumping. By applying a pump tone to excite propagating spin waves in a yttrium iron garnet delay line, four-wave mixing converts the pump mode into two phase-autonomous propagating magnon modes, i.e. a spontaneous mode with nearly twice the wavenumber of the pump mode and an idler mode with nearly zero wavenumber. This allows us to reliably generate ultrasharp spin wave dynamics with broad frequency tunability from the pump and magnetic field. We show that the spontaneous mode can be phase-locked to a probe tone, similar to an auto-oscillator. Furthermore, the spontaneous dynamics can be used to implement a high-gain magnonic parametric amplifier with a gain up to 40 dB. Our results open a new avenue for studying nonlinear magnonics and synchronization physics in propagating magnon geometry and for developing new magnonic devices.

cond-mat.mes-hall↗

arXiv2026

Non-Abelian Aharonov-Bohm Caging in Synthetic Dimensions with a Trapped Ion

Aharonov-Bohm (AB) caging is a complete localization phenomenon in two-dimensional lattices due to destructive interference induced by the background gauge fields. However, current investigations of AB caging are mostly restricted to the Abelian gauge field case, and the observation of AB caging under non-Abelian gauge fields in a quantum system still remains elusive. Here, we report experimental realization of tunable synthetic non-Abelian SU(2) gauge fields in a rhombic lattice, engineered within the synthetic dimensions of a vibrating trapped ion with multiple levels. We realize AB caging under both Abelian and non-Abelian gauge fields and systematically investigate the distinctive transport properties of the non-Abelian case. In particular, we observe typical emergent quantum dynamics unique to non-Abelian AB caging, including initial-state-dependent dynamics, second-order effects, and asymmetric caging behavior. These observations demonstrate the trapped ion system as a powerful platform for simulating emergent phenomena in high-dimensional quantum systems with exotic synthetic gauge fields.

quant-ph↗

arXiv2026

GPR: Towards a Generative Pre-trained One-Model Paradigm for Large-Scale Advertising Recommendation

As an intelligent infrastructure connecting users with commercial content, advertising recommendation systems play a central role in information flow and value creation within the digital economy. However, existing multi-stage advertising recommendation systems suffer from objective misalignment and error propagation, making it difficult to achieve global optimality, while unified generative recommendation models still struggle to meet the demands of practical industrial applications. To address these issues, we propose GPR (Generative Pre-trained Recommender), the first one-model framework that redefines advertising recommendation as an end-to-end generative task, replacing the traditional cascading paradigm with a unified generative approach. To realize GPR, we introduce three key innovations spanning unified representation, network architecture, and training strategy. First, we design a unified input schema and tokenization method tailored to advertising scenarios, mapping both ads and organic content into a shared multi-level semantic ID space, thereby enhancing semantic alignment and modeling consistency across heterogeneous data. Second, we develop the Heterogeneous Hierarchical Decoder (HHD), a dual-decoder architecture that decouples user intent modeling from ad generation, achieving a balance between training efficiency and inference flexibility while maintaining strong modeling capacity. Finally, we propose a multi-stage joint training strategy that integrates Multi-Token Prediction (MTP), Value-Aware Fine-Tuning and the Hierarchy Enhanced Policy Optimization (HEPO) algorithm, forming a complete generative recommendation pipeline that unifies interest modeling, value alignment, and policy optimization. GPR has been fully deployed in the Tencent Weixin Channels advertising system, delivering significant improvements in key business metrics including GMV and CTCVR.

cs.IR↗