Searcharxiv⌕ Search

arXiv subjects

Zhiyao Zhang

Publications and source records attributed to Zhiyao Zhang.

At least 19 recordsLinked to original sources
arXiv2026

GRF-Recon: Global Ray-Field Optimization for Long-Sequence Feed-forward Reconstruction

Feed-forward 3D reconstruction provides an efficient paradigm for scene modeling from image sequences. Scaling these models to large monocular scenarios are constrained by excessive GPU memory footprint, degraded local geometry, and long-term trajectory drift. Existing chunk-based optimization strategies provide limited geometric constraints and fail to maintain global consistency over extended trajectories. We present a unified framework for stable and scalable feed-forward 3D reconstruction from long monocular sequences. Our approach builds on coarse-to-fine trajectory alignment augmented by lightweight geometric prior injection. Distilling monocular geometric cues into the feed-forward backbone via LoRA adaptation improves depth accuracy on fine structures while preserving inference efficiency. We introduce a hybrid-weight sparse ray-field optimization that leverages high-frequency geometric features to guide local point-cloud refinement and enforce consistent inter-frame ray constraints. Unlike prior chunk-based methods, this establishes strong cross-frame geometric coupling while maintaining scalability. Finally, an efficient trajectory stitching strategy with joint ray-error optimization explicitly reduces accumulated drift. Extensive experiments show that our approach achieves competitive trajectory accuracy compared with representative SLAM systems, while maintaining globally consistent 3D reconstruction in large-scale scenarios.

cs.CV↗
arXiv2026

Existence of minimal models for threefold generalized pairs in positive characteristic

Let $\mathbb{K}$ be an algebraically closed field of characteristic $p>5$. We show the existence of minimal models for pseudo-effective NQC lc generalized pairs in dimension three over $\mathbb{K}$. As a consequence, we prove the termination of flips for pseudo-effective threefold NQC lc generalized pairs over $\mathbb{K}$. This provides a new proof on the termination of flips for pseudo-effective pairs over $\mathbb{K}$ without using the non-vanishing theorems. A key ingredient of our proof is the ACC for lc thresholds in dimension $\leq 3$ and the global ACC in dimension $\leq 2$ for generalized pairs over $\mathbb{K}$.

math.AG↗
arXiv2026

Semantics-Aware Bilevel Co-Evolution: Towards Automated Multicomponent Algorithm Design

LLM-assisted evolutionary search (LES) has emerged as a promising paradigm for automated algorithm design. However, existing methods usually suffer from two inherent limitations when facing the automated design of real-world complex algorithms that usually consist of multiple components. The first limitation is that they either focus on modifying entire algorithms, making it difficult to reuse high-quality components, or concentrate on component refinement within a limited set of predefined multicomponent configurations. The second limitation is the insufficient explicit modeling and exploitation of algorithm semantics. These limitations severely degrade search efficiency and hinder effective exploration of complex design spaces. Therefore, this paper proposes STABLE (Semantics-Aware Bilevel Co-Evolution), an LES method purpose-built for automated multicomponent algorithm design that introduces structural algorithm formulation and semantics-driven evolution. In STABLE, complex algorithms are organized into hierarchical and modular architectures rooted in domain knowledge, aligning the search space with their intrinsic compositional traits. Based on this structured algorithm formulation, STABLE simultaneously optimizes high-level multicomponent configurations and low-level functional components, enabling coordinated cross-level updates while maintaining suitable granularities for design space exploration. At each level, STABLE establishes a multi-faceted semantic model to assist LLMs in capturing structural correlations, functional compatibilities, and inherent rationalities among algorithm components. This semantic model serves as the core guidance for evolutionary search, enabling principled algorithm generation and algorithm evaluation. Extensive experiments demonstrate that STABLE outperform both human-designed baselines and those from advanced LES methods.

cs.NE↗
arXiv2026

Enabling Real-Time Phase Control in Traffic Signal Hardware-in-the-Loop Simulation

Advanced Traffic Signal Control (TSC) algorithms require real-time phase control, yet existing Hardware-in-the-Loop Simulation (HILS) testbeds only support pre-programmed timing plans. In this paper, we present the first HILS testbed for real-time phase control. We develop a novel middleware architecture that translates dynamic phase actions (selection, switch, and duration) into commands for NTCIP-compliant commercial hardware controllers. This middleware manages phase transitions, synchronizes signal states, and handles errors without interrupting the hardware's internal operations. Experimental validation demonstrates that the system executes real-time phase commands, handles system conflicts, and achieves a low system internal latency at sub-millisecond on average.

eess.SY↗
arXiv2026

A Tale of Two Problems: Multi-Task Bilevel Learning Meets Equality Constrained Multi-Objective Optimization

In recent years, bilevel optimization (BLO) has attracted significant attention for its broad applications in machine learning. However, most existing works on BLO remain confined to the single-task setting and rely on the lower-level strong convexity assumption, which significantly restricts their applicability to modern machine learning problems of growing complexity. In this paper, we make the first attempt to extend BLO to the multi-task setting under a relaxed lower-level general convexity (LLGC) assumption. To this end, we reformulate the multi-task bilevel learning (MTBL) problem with LLGC into an equality constrained multi-objective optimization (ECMO) problem. However, ECMO itself is a new problem that has not yet been studied in the literature. To address this gap, we first establish a new Karush-Kuhn-Tucker (KKT)-based Pareto stationarity as the convergence criterion for ECMO algorithm design. Based on this foundation, we propose a weighted Chebyshev (WC)-penalty algorithm that achieves a finite-time convergence rate of $O(ST^{-\frac{1}{2})$ to KKT-based Pareto stationarity in both deterministic and stochastic settings, where $S$ denotes the number of objectives, and $T$ is the total iterations. Moreover, by varying the preference vector over the $S$-dimensional simplex, our WC-penalty method systematically explores the Pareto front. Finally, solutions to the ECMO problem translate directly into solutions for the original MTBL problem, thereby closing the loop between these two foundational optimization frameworks.

cs.LG↗
arXiv2026

Improving bounds for value sets of polynomials over finite fields

Let $\mathbb{F}_{q}$ be a finite field of characteristic $p$, and let $f \in \mathbb{F}_{q}[x]$ be a polynomial of degree $d > 0$. Denote the image set of this polynomial as $V_{f}=\{f(α)\midα\in\mathbb{F}_{q}\}$ and denote the cardinality of this set as $N_{f}$. A much sharper bound for $N_{f}$ is established in this paper. In particular, for any $p\neq 2, 3$, and for nearly every generic quartic polynomial $f \in \mathbb{F}_{q}[x]$, we obtain $$\lvert N_f - \frac{5}{8} q \rvert \leq \frac{1}{2}\sqrt{q} + \frac{15}{4},$$ which holds as a simple corollary of the main result.

math.NT↗
arXiv2026

Characterising Toxicity in Generative Large Language Models

In recent years, the advent of the attention mechanism has significantly advanced the field of natural language processing (NLP), revolutionizing text processing and text generation. This has come about through transformer-based decoder-only architectures, which have become ubiquitous in NLP due to their impressive text processing and generation capabilities. Despite these breakthroughs, language models (LMs) remain susceptible to generating undesired outputs: inappropriate, offensive, or otherwise harmful responses. We will collectively refer to these as ``toxic'' outputs. Although methods like reinforcement learning from human feedback (RLHF) have been developed to align model outputs with human values, these safeguards can often be circumvented through carefully crafted prompts. Therefore, this paper examines the extent to which LLMs generate toxic content when prompted, as well as the linguistic factors -- both lexical and syntactic -- that influence the production of such outputs in generative models.

cs.CL↗
arXiv2025

Multi-Objective Bilevel Learning

As machine learning (ML) applications grow increasingly complex in recent years, modern ML frameworks often need to address multiple potentially conflicting objectives with coupled decision variables across different layers. This creates a compelling need for multi-objective bilevel learning (MOBL). So far, however, the field of MOBL remains in its infancy and many important problems remain under-explored. This motivates us to fill this gap and systematically investigate the theoretical and algorithmic foundation of MOBL. Specifically, we consider MOBL problems with multiple conflicting objectives guided by preferences at the upper-level subproblem, where part of the inputs depend on the optimal solution of the lower-level subproblem. Our goal is to develop efficient MOBL optimization algorithms to (1) identify a preference-guided Pareto-stationary solution with low oracle complexity; and (2) enable systematic Pareto front exploration. To this end, we propose a unifying algorithmic framework called weighted-Chebyshev multi-hyper-gradient-descent (WC-MHGD) for both deterministic and stochastic settings with finite-time Pareto-stationarity convergence rate guarantees, which not only implies low oracle complexity but also induces systematic Pareto front exploration. We further conduct extensive experiments to confirm our theoretical results.

cs.LG↗
arXiv2025

Finite-Time Global Optimality Convergence in Deep Neural Actor-Critic Methods for Decentralized Multi-Agent Reinforcement Learning

Actor-critic methods for decentralized multi-agent reinforcement learning (MARL) facilitate collaborative optimal decision making without centralized coordination, thus enabling a wide range of applications in practice. To date, however, most theoretical convergence studies for existing actor-critic decentralized MARL methods are limited to the guarantee of a stationary solution under the linear function approximation. This leaves a significant gap between the highly successful use of deep neural actor-critic for decentralized MARL in practice and the current theoretical understanding. To bridge this gap, in this paper, we make the first attempt to develop a deep neural actor-critic method for decentralized MARL, where both the actor and critic components are inherently non-linear. We show that our proposed method enjoys a global optimality guarantee with a finite-time convergence rate of O(1/T), where T is the total iteration times. This marks the first global convergence result for deep neural actor-critic methods in the MARL literature. We also conduct extensive numerical experiments, which verify our theoretical results.

cs.LG↗
arXiv2025

Consensus-based Decentralized Multi-agent Reinforcement Learning for Random Access Network Optimization

With wireless devices increasingly forming a unified smart network for seamless, user-friendly operations, random access (RA) medium access control (MAC) design is considered a key solution for handling unpredictable data traffic from multiple terminals. However, it remains challenging to design an effective RA-based MAC protocol to minimize collisions and ensure transmission fairness across the devices. While existing multi-agent reinforcement learning (MARL) approaches with centralized training and decentralized execution (CTDE) have been proposed to optimize RA performance, their reliance on centralized training and the significant overhead required for information collection can make real-world applications unrealistic. In this work, we adopt a fully decentralized MARL architecture, where policy learning does not rely on centralized tasks but leverages consensus-based information exchanges across devices. We design our MARL algorithm over an actor-critic (AC) network and propose exchanging only local rewards to minimize communication overhead. Furthermore, we provide a theoretical proof of global convergence for our approach. Numerical experiments show that our proposed MARL algorithm can significantly improve RA network performance compared to other baselines.

cs.NI↗
arXiv2025

InstanceGaussian: Appearance-Semantic Joint Gaussian Representation for 3D Instance-Level Perception

3D scene understanding has become an essential area of research with applications in autonomous driving, robotics, and augmented reality. Recently, 3D Gaussian Splatting (3DGS) has emerged as a powerful approach, combining explicit modeling with neural adaptability to provide efficient and detailed scene representations. However, three major challenges remain in leveraging 3DGS for scene understanding: 1) an imbalance between appearance and semantics, where dense Gaussian usage for fine-grained texture modeling does not align with the minimal requirements for semantic attributes; 2) inconsistencies between appearance and semantics, as purely appearance-based Gaussians often misrepresent object boundaries; and 3) reliance on top-down instance segmentation methods, which struggle with uneven category distributions, leading to over- or under-segmentation. In this work, we propose InstanceGaussian, a method that jointly learns appearance and semantic features while adaptively aggregating instances. Our contributions include: i) a novel Semantic-Scaffold-GS representation balancing appearance and semantics to improve feature representations and boundary delineation; ii) a progressive appearance-semantic joint training strategy to enhance stability and segmentation accuracy; and iii) a bottom-up, category-agnostic instance aggregation approach that addresses segmentation challenges through farthest point sampling and connected component analysis. Our approach achieves state-of-the-art performance in category-agnostic, open-vocabulary 3D point-level segmentation, highlighting the effectiveness of the proposed representation and training strategies. Project page: https://lhj-git.github.io/InstanceGaussian/

cs.CV↗
arXiv2025

Real-World Deployment and Assessment of a Multi-Agent Reinforcement Learning-Based Variable Speed Limit Control System

This article presents the first field deployment of a multi-agent reinforcement learning (MARL) based variable speed limit (VSL) control system on Interstate 24 (I-24) near Nashville, Tennessee. We design and demonstrate a full pipeline from training MARL agents in a traffic simulator to a field deployment on a 17-mile segment of I-24 encompassing 67 VSL controllers. The system was launched on March 8th, 2024, and has made approximately 35 million decisions on 28 million trips in six months of operation. We apply an invalid action masking mechanism and several safety guards to ensure real-world constraints. The MARL-based implementation operates up to 98% of the time, with the safety guards overriding the MARL decisions for the remaining time. We evaluate the performance of the MARL-based algorithm in comparison to a previously deployed non-RL VSL benchmark algorithm on I-24. Results show that the MARL-based VSL control system achieves a superior performance. The accuracy of correctly warning drivers about slowing traffic ahead is improved by 14% and the response delay to non-recurrent congestion is reduced by 75%. The preliminary data shows that the VSL control system has reduced the crash rate by 26% and the secondary crash rate by 50%. We open-sourced the deployed MARL-based VSL algorithm at https://github.com/Lab-Work/marl-vsl-controller.

eess.SY↗
arXiv2024

Prediction of three-dimensional chemically reacting compressible turbulence based on implicit U-Net enhanced Fourier neural operator

The accurate and fast prediction of long-term dynamics of turbulence presents a significant challenge for both traditional numerical simulations and machine learning methods. In recent years, the emergence of neural operators has provided a promising approach to address this issue. The implicit U-Net enhanced Fourier neural operator (IU-FNO) has successfully demonstrated long-term stable predictions for three-dimensional incompressible turbulence. In this study, we extend this method to the three-dimensional chemically reacting compressible turbulence. Numerical results show that the IU-FNO model predicts flow dynamics significantly faster than the traditional dynamic Smagorinsky model (DSM) used in large eddy simulation (LES). In terms of prediction accuracy, the IU-FNO framework outperforms the traditional DSM in predicting the energy spectra of velocity, temperature, and density, the probability density functions (PDFs) of vorticity and velocity increments, and instantaneous spatial structures of temperature. Therefore, the IU-FNO represents a highly promising approach for predicting chemically reacting compressible turbulence.

physics.flu-dyn↗
arXiv2024

Phase Re-service in Reinforcement Learning Traffic Signal Control

This article proposes a novel approach to traffic signal control that combines phase re-service with reinforcement learning (RL). The RL agent directly determines the duration of the next phase in a pre-defined sequence. Before the RL agent's decision is executed, we use the shock wave theory to estimate queue expansion at the designated movement allowed for re-service and decide if phase re-service is necessary. If necessary, a temporary phase re-service is inserted before the next regular phase. We formulate the RL problem as a semi-Markov decision process (SMDP) and solve it with proximal policy optimization (PPO). We conducted a series of experiments that showed significant improvements thanks to the introduction of phase re-service. Vehicle delays are reduced by up to 29.95% of the average and up to 59.21% of the standard deviation. The number of stops is reduced by 26.05% on average with 45.77% less standard deviation.

eess.SY↗
arXiv2024

Field Deployment of Multi-Agent Reinforcement Learning Based Variable Speed Limit Controllers

This article presents the first field deployment of a multi-agent reinforcement-learning (MARL) based variable speed limit (VSL) control system on the I-24 freeway near Nashville, Tennessee. We describe how we train MARL agents in a traffic simulator and directly deploy the simulation-based policy on a 17-mile stretch of Interstate 24 with 67 VSL controllers. We use invalid action masking and several safety guards to ensure the posted speed limits satisfy the real-world constraints from the traffic management center and the Tennessee Department of Transportation. Since the time of launch of the system through April, 2024, the system has made approximately 10,000,000 decisions on 8,000,000 trips. The analysis of the controller shows that the MARL policy takes control for up to 98% of the time without intervention from safety guards. The time-space diagrams of traffic speed and control commands illustrate how the algorithm behaves during rush hour. Finally, we quantify the domain mismatch between the simulation and real-world data and demonstrate the robustness of the MARL policy to this mismatch.

cs.MA↗
arXiv2024

Broadband microwave waveform generation with programmable chirp shapes via recirculating phase-modulated optical fiber loop controlled by low-speed electronics

Broadband microwave waveforms with programmable chirp shapes are captivating in numerous practical applications. Compared with electronic technology, photonic-assisted solutions exhibit excellent performance in bandwidth and flexibility, but still suffer from complex architecture and requirement of high-speed electronics. Besides, rapid manipulation of chirp shape is still a challenge in the scientific community. In this paper, we propose and demonstrate a novel concept for generating broadband microwave waveforms with programmable chirp shapes. This concept is realized on a simple fiber-optic platform involving a continuous-wave laser source, a recirculating phase-modulated optical fiber loop, and low-speed electronics with a sampling rate at the level of MS/s. Based on this method, chirped microwave waveforms with a bandwidth up to tens of GHz can be generated, where the chirp shape is identical to the low-frequency driving waveform of the recirculating phase-modulated optical fiber loop. In addition, all the parameters of the generated chirped microwave waveforms can be easily reconfigured in real time, including the bandwidth, the central frequency, and the temporal duration. In the experiment, broadband microwave waveforms with customized chirp shapes are generated, where the center frequency and bandwidth tuning ranges exceed 21 GHz, the temporal duration is tuned in the range of 9 ns to 180 ns, and the coherent time of the generated microwave waveform is larger than 100 μs. This simple fiber-optic platform paves a way to generate broadband microwave waveforms with user-definable chirp shapes, which can find applications in broadband radar systems, electronic warfare and wireless communications.

physics.optics↗
arXiv2024

S3-SLAM: Sparse Tri-plane Encoding for Neural Implicit SLAM

With the emergence of Neural Radiance Fields (NeRF), neural implicit representations have gained widespread applications across various domains, including simultaneous localization and mapping. However, current neural implicit SLAM faces a challenging trade-off problem between performance and the number of parameters. To address this problem, we propose sparse tri-plane encoding, which efficiently achieves scene reconstruction at resolutions up to 512 using only 2~4% of the commonly used tri-plane parameters (reduced from 100MB to 2~4MB). On this basis, we design S3-SLAM to achieve rapid and high-quality tracking and mapping through sparsifying plane parameters and integrating orthogonal features of tri-plane. Furthermore, we develop hierarchical bundle adjustment to achieve globally consistent geometric structures and reconstruct high-resolution appearance. Experimental results demonstrate that our approach achieves competitive tracking and scene reconstruction with minimal parameters on three datasets. Source code will soon be available.

cs.CV↗