arXiv subjects

Li Chen

Publications and source records attributed to Li Chen.

At least 181 records · Page 10Linked to original sources

arXiv2025

XVerse: Consistent Multi-Subject Control of Identity and Semantic Attributes via DiT Modulation

Achieving fine-grained control over subject identity and semantic attributes (pose, style, lighting) in text-to-image generation, particularly for multiple subjects, often undermines the editability and coherence of Diffusion Transformers (DiTs). Many approaches introduce artifacts or suffer from attribute entanglement. To overcome these challenges, we propose a novel multi-subject controlled generation model XVerse. By transforming reference images into offsets for token-specific text-stream modulation, XVerse allows for precise and independent control for specific subject without disrupting image latents or features. Consequently, XVerse offers high-fidelity, editable multi-subject image synthesis with robust control over individual subject characteristics and semantic attributes. This advancement significantly improves personalized and complex scene generation capabilities.

cs.CV↗

arXiv2025

Single-Agent vs. Multi-Agent LLM Strategies for Automated Student Reflection Assessment

We explore the use of Large Language Models (LLMs) for automated assessment of open-text student reflections and prediction of academic performance. Traditional methods for evaluating reflections are time-consuming and may not scale effectively in educational settings. In this work, we employ LLMs to transform student reflections into quantitative scores using two assessment strategies (single-agent and multi-agent) and two prompting techniques (zero-shot and few-shot). Our experiments, conducted on a dataset of 5,278 reflections from 377 students over three academic terms, demonstrate that the single-agent with few-shot strategy achieves the highest match rate with human evaluations. Furthermore, models utilizing LLM-assessed reflection scores outperform baselines in both at-risk student identification and grade prediction tasks. These findings suggest that LLMs can effectively automate reflection assessment, reduce educators' workload, and enable timely support for students who may need additional assistance. Our work emphasizes the potential of integrating advanced generative AI technologies into educational practices to enhance student engagement and academic success.

cs.LG↗

arXiv2025

Mixed-Precision Quantization: Make the Best Use of Bits Where They Matter Most

Mixed-precision quantization offers superior performance to fixed-precision quantization. It has been widely used in signal processing, communication systems, and machine learning. In mixed-precision quantization, bit allocation is essential. Hence, in this paper, we propose a new bit allocation framework for mixed-precision quantization from a search perspective. First, we formulate a general bit allocation problem for mixed-precision quantization. Then we introduce the penalized particle swarm optimization (PPSO) algorithm to address the integer consumption constraint. To improve efficiency and avoid iterations on infeasible solutions within the PPSO algorithm, a greedy criterion particle swarm optimization (GC-PSO) algorithm is proposed. The corresponding convergence analysis is derived based on dynamical system theory. Furthermore, we apply the above framework to some specific classic fields, i.e., finite impulse response (FIR) filters, receivers, and gradient descent. Numerical examples in each application underscore the superiority of the proposed framework to the existing algorithms.

eess.SP↗

arXiv2025

Revisiting MUSIC: A Finite-Precision Perspective

The high computational complexity of the multiple signal classification (MUSIC) algorithm is mainly caused by the subspace decomposition and spectrum search, especially for frequent real-time applications or massive sensors. In this paper, we propose a low-complexity MUSIC algorithm from a finite-precision arithmetic perspective. First, we analyze the computational bottlenecks of the classic low-complexity randomized unitary-based MUSIC (RU-MUSIC), formulating this computational issue as an inner product problem. Then, a mixed-precision method is introduced to address this problem. Specifically, this method partitions summations in inner products into blocks, where intra-block computations use low-precision arithmetic and inter-block sums use high-precision arithmetic. To further improve computational accuracy, we develop an adaptive-precision method that supports adaptive block sizes and multiple precision levels. Finally, simulation results show that the proposed finite-precision MUSIC design achieves direction-of-arrival (DOA) estimation performance similar to that using full-precision arithmetic while reducing more than 50\% computational cost.

eess.SP↗

arXiv2025

Reinforced Refinement with Self-Aware Expansion for End-to-End Autonomous Driving

End-to-end autonomous driving has emerged as a promising paradigm for directly mapping sensor inputs to planning maneuvers using learning-based modular integrations. However, existing imitation learning (IL)-based models suffer from generalization to hard cases, and a lack of corrective feedback loop under post-deployment. While reinforcement learning (RL) offers a potential solution to tackle hard cases with optimality, it is often hindered by overfitting to specific driving cases, resulting in catastrophic forgetting of generalizable knowledge and sample inefficiency. To overcome these challenges, we propose Reinforced Refinement with Self-aware Expansion (R2SE), a novel learning pipeline that constantly refines hard domain while keeping generalizable driving policy for model-agnostic end-to-end driving systems. Through reinforcement fine-tuning and policy expansion that facilitates continuous improvement, R2SE features three key components: 1) Generalist Pretraining with hard-case allocation trains a generalist imitation learning (IL) driving system while dynamically identifying failure-prone cases for targeted refinement; 2) Residual Reinforced Specialist Fine-tuning optimizes residual corrections using reinforcement learning (RL) to improve performance in hard case domain while preserving global driving knowledge; 3) Self-aware Adapter Expansion dynamically integrates specialist policies back into the generalist model, enhancing continuous performance improvement. Experimental results in closed-loop simulation and real-world datasets demonstrate improvements in generalization, safety, and long-horizon policy robustness over state-of-the-art E2E systems, highlighting the effectiveness of reinforce refinement for scalable autonomous driving.

cs.RO↗

arXiv2025

Causality-Inspired Fair Representation Learning for Multimodal Recommendation

Recently, multimodal recommendations (MMR) have gained increasing attention for alleviating the data sparsity problem of traditional recommender systems by incorporating modality-based representations. Although MMR exhibits notable improvement in recommendation accuracy, we empirically validate that an increase in the quantity or variety of modalities leads to a higher degree of users' sensitive information leakage due to entangled causal relationships, risking fair representation learning. On the other hand, existing fair representation learning approaches are mostly based on the assumption that sensitive information is solely leaked from users' interaction data and do not explicitly model the causal relationships introduced by multimodal data, which limits their applicability in multimodal scenarios. To address this limitation, we propose a novel fair multimodal recommendation approach (dubbed FMMRec) through causality-inspired fairness-oriented modal disentanglement and relation-aware fairness learning. Particularly, we disentangle biased and filtered modal embeddings inspired by causal inference techniques, enabling the mining of modality-based unfair and fair user-user relations, thereby enhancing the fairness and informativeness of user representations. By addressing the causal effects of sensitive attributes on user preferences, our approach aims to achieve counterfactual fairness in multimodal recommendations. Experiments on two public datasets demonstrate the superiority of our FMMRec relative to the state-of-the-art baselines. Our source code is available at https://github.com/WeixinChen98/FMMRec.

cs.IR↗

arXiv2025

Improved adjoint lattice Boltzmann method for topology optimization of laminar convective heat transfer

Solving flow-related inverse problems such as topology optimization problems is intricate but significant in various engineering fields. The lattice Boltzmann method (LBM) and the related adjoint method are highly suitable to perform sensitivity analysis in flow-related inverse problems thanks to their strong capability to handle complex structures and excellent parallel scalability. However, the current continuous adjoint LBM shows theoretical inconsistency and poor numerical stability for open flow systems. To solve these issues, the present work develops the fully consistent adjoint boundary conditions from the discrete adjoint LBM. For the first time, the gap between the two adjoint LBMs is unveiled by rigorously deriving both the continuous and discrete adjoint LBMs and comprehensively evaluating their numerical performances in the 2D and 3D pipe bend optimization cases. It is revealed that theoretical inconsistency or singularity exists in the continuous adjoint boundary conditions for open flow systems, corresponding to a much inferior numerical stability of the adjoint solution and an obvious numerical error in sensitivity. Fully consistent adjoint boundary conditions in elegant local form are derived from the discrete adjoint LBM in this work, which can always acquire exact sensitivity results and the theoretically highest numerical stability, with a 10 times higher Reynolds number (Re) achieved while without any increase of computational cost. 3D microchannel heat sinks under various Re are designed, and the esthetic and physically reasonable optimized designs are obtained under various parameter settings, demonstrating the necessity and versatility of the presented discrete adjoint LBM.

math.NA↗

arXiv2025

Exploring bidirectional bounds for minimax-training of Energy-based models

Energy-based models (EBMs) estimate unnormalized densities in an elegant framework, but they are generally difficult to train. Recent work has linked EBMs to generative adversarial networks, by noting that they can be trained through a minimax game using a variational lower bound. To avoid the instabilities caused by minimizing a lower bound, we propose to instead work with bidirectional bounds, meaning that we maximize a lower bound and minimize an upper bound when training the EBM. We investigate four different bounds on the log-likelihood derived from different perspectives. We derive lower bounds based on the singular values of the generator Jacobian and on mutual information. To upper bound the negative log-likelihood, we consider a gradient penalty-like bound, as well as one based on diffusion processes. In all cases, we provide algorithms for evaluating the bounds. We compare the different bounds to investigate, the pros and cons of the different approaches. Finally, we demonstrate that the use of bidirectional bounds stabilizes EBM training and yields high-quality density estimation and sample generation.

cs.LG↗

arXiv2025

Two-dimensional signal-dependent parabolic-elliptic Keller-Segel system and its mean-field derivation

In this paper, the well-posedness of two-dimensional signal-dependent Keller-Segel system and its mean-field derivation from a interacting particle system on the whole space are investigated. The signal dependence effect is reflected by the fact that the diffusion coefficient in the particle system depends non-linearly on the interactions between the individuals. Therefore, the mathematical challenge in studying the well-posedness of this system lies in the possible degeneracy and the aggregation effect when the concentration of signal becomes unbounded. The well-established method on bounded domains, to obtain the appropriate estimates for the signal concentration, is invalid for the whole space case. Motivated by the entropy minimization method and Onofri's inequality, which has been successfully applied for the parabolic-parabolic Keller-Segel system, we establish a complete entropy estimate benefited from the linear diffusion term, which plays an important role in obtaining the L^p estimates for the solution. Furthermore, the upper bound for the concentration of signal is obtained. Based on the estimates we obtained for the density of bacteria, the rigorous mean-field derivation is proved by introducing an intermediate particle system with a mollified interaction potential with logarithmic scaling. By using this mollification, we obtain the convergence of the particle trajectories in expectation, which implies the weak propagation of chaos. Additionally, under a regularity assumption of the initial data, we infer higher regularity for the solutions, which allows us to use the relative entropy method to derive the strong L^1 convergence for the propagation of chaos.

math.AP↗

arXiv2025

Resource Heterogeneity-Aware and Utilization-Enhanced Scheduling for Deep Learning Clusters

Scheduling deep learning (DL) models to train on powerful clusters with accelerators like GPUs and TPUs, presently falls short, either lacking fine-grained heterogeneity awareness or leaving resources substantially under-utilized. To fill this gap, we propose a novel design of a task-level heterogeneity-aware scheduler, Hadar, based on an optimization framework that can boost resource utilization. Hadar leverages the performance traits of DL jobs on a heterogeneous DL cluster, characterizes the task-level performance heterogeneity in the optimization problem, and makes scheduling decisions across both spatial and temporal dimensions. It involves the primal-dual framework employing a dual subroutine, to solve the optimization problem and guide the scheduling design. Our trace-driven simulation with representative DL model training workloads demonstrates that Hadar accelerates the total time duration by 1.20x when compared with its state-of-the-art heterogeneity-aware counterpart, Gavel. Further, our Hadar scheduler is enhanced to HadarE by forking each job into multiple copies to let a job train concurrently on heterogeneous GPUs resided on separate available nodes (i.e., machines or servers) for resource utilization enhancement. HadarE is evaluated extensively on physical DL clusters for comparison with Hadar and Gavel. With substantial enhancement in cluster resource utilization (by 1.45x), HadarE exhibits considerable speed-ups in DL model training, reducing the total time duration by 50% (or 80%) on an Amazon's AWS (or our lab) cluster, while producing trained DL models with consistently better inference quality than those trained by Hadar.

cs.DC↗

arXiv2025

Selected open cluster sample for validating atmospheric parameters: Application to Gaia and other surveys

Reliable stellar atmospheric parameters are essential for probing stellar structure and evolution, and for stellar population studies. However, various deviations appear in comparisons with different ground-based spectroscopic surveys. We aim to select high-quality open cluster members and employ the atmospheric parameters provided by the theoretical isochrones of open clusters as a benchmark to assess the quality of stellar atmospheric parameters from Gaia DR3 and other ground-based spectroscopic surveys, such as LAMOST DR11, APOGEE DR17, and GALAH DR4. We selected 130 open clusters with well-defined main sequences within 500 pc of the solar neighborhood as a benchmark sample to estimate the reference atmospheric parameters of the members from the best-fit isochrones of those clusters. By comparing the atmospheric parameters provided by different spectroscopic surveys to the theoretical parameters, we found that the atmospheric parameter deviation and the corresponding dispersions exhibit different variations. The atmospheric parameter deviations of F, G, and K-type stars are smaller than those of B, A, and M-type stars for most surveys. For most samples, the dispersion of Teff decreases as temperature decreases, whereas the dispersion of logg shows the opposite trend.

astro-ph.SR↗

arXiv2025

On the mean-field limit of Vlasov-Poisson-Fokker-Planck equations

The derivation of effective descriptions for interacting many-body systems is an important branch of applied mathematics. We prove a propagation of chaos result for a system of $N$ particles subject to Newtonian time evolution with or without additional white noise influencing the velocities of the particles. We assume that the particles interact according to a regularized Coulomb-interaction with a regularization parameter that vanishes in the $N\to\infty$ limit. The respective effective description is the so called Vlasov-Poisson-Fokker-Planck (VPFP), respectively the Vlasov-Poisson (VP) equation in the case of no or sub-dominant white noise. To obtain our result we combine the relative entropy method from \cite{jabinWang2016} with the control on the difference between the trajectories of the true and the effective description provided in \cite{HLP20} for the VPFP case respectively in \cite{LP} for the VP case. This allows us to prove strong convergence of the marginals, i.e. convergence in $L^1$.

math-ph↗

arXiv2025

On the Power of Learning-Augmented Search Trees

We study learning-augmented binary search trees (BSTs) via Treaps with carefully designed priorities. The result is a simple search tree in which the depth of each item $x$ is determined by its predicted weight $w_x$. Specifically, each item $x$ is assigned a composite priority of $-\lfloor\log\log(1/w_x)\rfloor + U(0, 1)$ where $U(0, 1)$ is the uniform random variable. By choosing $w_x$ as the relative frequency of $x$, the resulting search trees achieve static optimality. This approach generalizes the recent learning-augmented BSTs [Lin-Luo-Woodruff ICML '22], which only work for Zipfian distributions, by extending them to arbitrary input distributions. Furthermore, we demonstrate that our method can be generalized to a B-Tree data structure using the B-Treap approach [Golovin ICALP '09]. Our search trees are also capable of leveraging localities in the access sequence through online self-reorganization, thereby achieving the working-set property. Additionally, they are robust to prediction errors and support dynamic operations, such as insertions, deletions, and prediction updates. We complement our analysis with an empirical study, demonstrating that our method outperforms prior work and classic data structures.

cs.DS↗

arXiv2025

Korevaar-Schoen and heat kernel characterizations of Sobolev and BV spaces on local trees

We study Sobolev and BV spaces on local trees which are metric spaces locally isometric to real trees. Such spaces are equipped with a Radon measure satisfying a locally uniform volume growth condition. Using the intrinsic geodesic structure, we define weak gradients and develop from it a coherent theory of Sobolev and BV spaces. We provide two main characterizations: one via Korevaar-Schoen-type energy functionals and another via the heat kernel associated with the natural Dirichlet form. Applications include interpolation results for Besov-Lipschitz spaces, critical exponents computations, and a Nash inequality. In globally tree-like settings we also establish $L^p$ gradient bounds for the heat semigroup.

math.AP↗

arXiv2025

FullStack Bench: Evaluating LLMs as Full Stack Coders

As the capabilities of code large language models (LLMs) continue to expand, their applications across diverse code intelligence domains are rapidly increasing. However, most existing datasets only evaluate limited application domains. To address this gap, we have developed a comprehensive code evaluation dataset FullStack Bench focusing on full-stack programming, which encompasses a wide range of application domains (e.g., basic programming, data analysis, software engineering, mathematics, and machine learning). Besides, to assess multilingual programming capabilities, in FullStack Bench, we design real-world instructions and corresponding unit test cases from 16 widely-used programming languages to reflect real-world usage scenarios rather than simple translations. Moreover, we also release an effective code sandbox execution tool (i.e., SandboxFusion) supporting various programming languages and packages to evaluate the performance of our FullStack Bench efficiently. Comprehensive experimental results on our FullStack Bench demonstrate the necessity and effectiveness of our FullStack Bench and SandboxFusion.

cs.AI↗

arXiv2025

Type-III Weyl Semi-Half-Metal in an Ultralight Monolayer Li$_2$N

The interplay between magnetic ordering and band topology has emerged as a fertile ground for discovering novel quantum states with profound implications for fundamental physics and next-generation electronics. Here, we theoretically predict a new type-III Weyl semi-half-metal (SHM) state in monolayer Li$_2$N, uniquely combining magnetic half-metallicity and type-III Weyl semimetal characteristics. First-principles calculations reveal a fully spin-polarized and critically tilted Weyl cone around the Fermi level in monolayer Li$_2$N, driven by $p$-orbital ferromagnetism. This arises from the symmetry-protected band crossing between a flat valence band and a highly dispersive conduction band, leading to type-III Weyl fermions with strong transport anisotropy. A low-energy $k{\cdot}p$ Hamiltonian is constructed and corresponding nontrivial edge states are uncovered to capture the topological nature of Li$_2$N. Notably, this Weyl SHM phase remains robust under biaxial strain ranging from -2$\%$ to $4\%$, with an ideal type-III Weyl fermion emerging alongside a line-like ergodic surface emerging at 3.7$\%$ strain, offering a promising platform for exploring correlated electronic phenomena. Our results establish Li$_2$N as a viable candidate for realizing exotic type-III Weyl SHM states and open a new avenue for exploring the intricate interplay among magnetism, topology, and flat-band physics.

cond-mat.mtrl-sci↗

arXiv2025

NN-Based Joint Mitigation of IQ Imbalance and PA Nonlinearity With Multiple States

Joint mitigation of IQ imbalance and PA nonlinearity is important for improving the performance of radio frequency (RF) transmitters. In this paper, we propose a new neural network (NN) model, which can be used for joint digital pre-distortion (DPD) of non-ideal IQ modulators and PAs in a transmitter with multiple operating states. The model is based on the methodology of multi-task learning (MTL). In this model, the hidden layers of the main NN are shared by all signal states, and the output layer's weights and biases are dynamically generated by another NN. The experimental results show that the proposed model can effectively perform joint DPD for IQ-PA systems, and it achieves better overall performance within multiple signal states than the existing methods.

eess.SY↗

arXiv2025

TigerVector: Supporting Vector Search in Graph Databases for Advanced RAGs

In this paper, we introduce TigerVector, a system that integrates vector search and graph query within TigerGraph, a Massively Parallel Processing (MPP) native graph database. We extend the vertex attribute type with the embedding type. To support fast vector search, we devise an MPP index framework that interoperates efficiently with the graph engine. The graph query language GSQL is enhanced to support vector type expressions and enable query compositions between vector search results and graph query blocks. These advancements elevate the expressive power and analytical capabilities of graph databases, enabling seamless fusion of unstructured and structured data in ways previously unattainable. Through extensive experiments, we demonstrate TigerVector's hybrid search capability, scalability, and superior performance compared to other graph databases (including Neo4j and Amazon Neptune) and a highly optimized specialized vector database (Milvus). TigerVector was integrated into TigerGraph v4.2, the latest release of TigerGraph, in December 2024.

cs.DB↗