arXiv subjects

Zekai Chen

Publications and source records attributed to Zekai Chen.

At least 55 records · Page 3Linked to original sources

arXiv2025

Rethinking Client-oriented Federated Graph Learning

As a new distributed graph learning paradigm, Federated Graph Learning (FGL) facilitates collaborative model training across local systems while preserving data privacy. We review existing FGL approaches and categorize their optimization mechanisms into: (1) Server-Client (S-C), where clients upload local model parameters for server-side aggregation and global updates; (2) Client-Client (C-C), which allows direct exchange of information between clients and customizing their local training process. We reveal that C-C shows superior potential due to its refined communication structure. However, existing C-C methods broadcast redundant node representations, incurring high communication costs and privacy risks at the node level. To this end, we propose FedC4, which combines graph Condensation with C-C Collaboration optimization. Specifically, FedC4 employs graph condensation technique to refine the knowledge of each client's graph into a few synthetic embeddings instead of transmitting node-level knowledge. Moreover, FedC4 introduces three novel modules that allow the source client to send distinct node representations tailored to the target client's graph properties. Experiments on eight public real-world datasets show that FedC4 outperforms state-of-the-art baselines in both task performance and communication cost. Our code is now available on https://github.com/Ereshkigal1/FedC4.

cs.LG↗

arXiv2025

SADA: Stability-guided Adaptive Diffusion Acceleration

Diffusion models have achieved remarkable success in generative tasks but suffer from high computational costs due to their iterative sampling process and quadratic attention costs. Existing training-free acceleration strategies that reduce per-step computation cost, while effectively reducing sampling time, demonstrate low faithfulness compared to the original baseline. We hypothesize that this fidelity gap arises because (a) different prompts correspond to varying denoising trajectory, and (b) such methods do not consider the underlying ODE formulation and its numerical solution. In this paper, we propose Stability-guided Adaptive Diffusion Acceleration (SADA), a novel paradigm that unifies step-wise and token-wise sparsity decisions via a single stability criterion to accelerate sampling of ODE-based generative models (Diffusion and Flow-matching). For (a), SADA adaptively allocates sparsity based on the sampling trajectory. For (b), SADA introduces principled approximation schemes that leverage the precise gradient information from the numerical ODE solver. Comprehensive evaluations on SD-2, SDXL, and Flux using both EDM and DPM++ solvers reveal consistent $\ge 1.8\times$ speedups with minimal fidelity degradation (LPIPS $\leq 0.10$ and FID $\leq 4.5$) compared to unmodified baselines, significantly outperforming prior methods. Moreover, SADA adapts seamlessly to other pipelines and modalities: It accelerates ControlNet without any modifications and speeds up MusicLDM by $1.8\times$ with $\sim 0.01$ spectrogram LPIPS.

cs.LG↗

arXiv2025

A Comprehensive Data-centric Overview of Federated Graph Learning

In the era of big data applications, Federated Graph Learning (FGL) has emerged as a prominent solution that reconcile the tradeoff between optimizing the collective intelligence between decentralized datasets holders and preserving sensitive information to maximum. Existing FGL surveys have contributed meaningfully but largely focus on integrating Federated Learning (FL) and Graph Machine Learning (GML), resulting in early stage taxonomies that emphasis on methodology and simulated scenarios. Notably, a data centric perspective, which systematically examines FGL methods through the lens of data properties and usage, remains unadapted to reorganize FGL research, yet it is critical to assess how FGL studies manage to tackle data centric constraints to enhance model performances. This survey propose a two-level data centric taxonomy: Data Characteristics, which categorizes studies based on the structural and distributional properties of datasets used in FGL, and Data Utilization, which analyzes the training procedures and techniques employed to overcome key data centric challenges. Each taxonomy level is defined by three orthogonal criteria, each representing a distinct data centric configuration. Beyond taxonomy, this survey examines FGL integration with Pretrained Large Models, showcases realistic applications, and highlights future direction aligned with emerging trends in GML.

cs.LG↗

arXiv2025

Terahertz time-domain spectroscopy of materials under high pressure in a diamond anvil cell

We present the combination of a broadband terahertz time-domain spectroscopy system (0.1 - 8 THz), a diamond anvil cell capable of generating high pressure conditions of up to 10 GPa and a cryostat reaching temperatures as low as 10 K. This combination allows us to perform equilibrium and time-resolved THz spectroscopy measurements of a sample while continuously tuning its temperature and pressure conditions. In this work, the procedures and characterizations necessary to carry out such experiments in a tabletop setup are presented. Due to the large modifications of the terahertz beam as it goes through the diamond anvil cell (DAC), standard terahertz time-domain spectroscopy analysis procedures are no longer applicable. New methods to extract the pressure dependent material parameters are presented, both for samples homogeneously filling the DAC sample chamber as well as for bulk samples embedded in pressure media. Different pressure media are tested and evaluated using these new methods, and the obtained material parameters are compared to literature values. Time resolved measurements under pressure are demonstrated using an optical pump - THz probe scheme.

cond-mat.mtrl-sci↗

arXiv2025

A Model Predictive Control Approach for Quadrotor Cruise Control

This paper investigates the application of a Model Predictive Controller (MPC) for the cruise control system of a quadrotor, focusing on hovering point stabilization and reference tracking. Initially, a full-state-feedback MPC is designed for the ideal scenario. To account for real-world conditions, a constant disturbance is introduced to the quadrotor, simulating a gust of wind in a specific direction. In response, an output-feedback offset-free MPC is developed to stabilize the quadrotor while rejecting the disturbance. We validate the design of the controller by conducting stability analysis, as well as numerical simulations under different circumstances. It is shown that the designed controller can achieve all the expected goals for the cruise control, including reference tracking and disturbance rejection. This project was implemented using Python and the CVXPY library for convex optimization.

eess.SY↗

arXiv2025

Naked Eye Three-dimensional Display System Based on Time-multiplexed Technology

Our group is developing a multi-user eye-tracked 3D display, an evolution of the single-user eye-tracked 3D display that we have already successfully developed. This display utilizes a slanted lenticular setup, where multiple perspective views are shown across the viewing field. Due to the constraints of the lenticular lens parameters, identical views are repeated across the field, limiting eye tracking to a single user. However, this limitation can be addressed using spatio-temporal multiplexing, where view zone groups are presented sequentially with a high frame rate liquid crystal display (LCD) and driver, in combination with a synchronized directional light emitting diode (LED) array. In this paper, we describe the operation and results of the backlight drive electronics, where a prototype using a white LED illumination matrix, a simplified LCD panel, and a linear Fresnel lens array serves as a test bed.

eess.SY↗

arXiv2024

FlexFL: Heterogeneous Federated Learning via APoZ-Guided Flexible Pruning in Uncertain Scenarios

Along with the increasing popularity of Deep Learning (DL) techniques, more and more Artificial Intelligence of Things (AIoT) systems are adopting federated learning (FL) to enable privacy-aware collaborative learning among AIoT devices. However, due to the inherent data and device heterogeneity issues, existing FL-based AIoT systems suffer from the model selection problem. Although various heterogeneous FL methods have been investigated to enable collaborative training among heterogeneous models, there is still a lack of i) wise heterogeneous model generation methods for devices, ii) consideration of uncertain factors, and iii) performance guarantee for large models, thus strongly limiting the overall FL performance. To address the above issues, this paper introduces a novel heterogeneous FL framework named FlexFL. By adopting our Average Percentage of Zeros (APoZ)-guided flexible pruning strategy, FlexFL can effectively derive best-fit models for heterogeneous devices to explore their greatest potential. Meanwhile, our proposed adaptive local pruning strategy allows AIoT devices to prune their received models according to their varying resources within uncertain scenarios. Moreover, based on self-knowledge distillation, FlexFL can enhance the inference performance of large models by learning knowledge from small models. Comprehensive experimental results show that, compared to state-of-the-art heterogeneous FL methods, FlexFL can significantly improve the overall inference accuracy by up to 14.24%.

cs.DC↗

arXiv2024

Advancing High Resolution Vision-Language Models in Biomedicine

Multi-modal learning has significantly advanced generative AI, especially in vision-language modeling. Innovations like GPT-4V and open-source projects such as LLaVA have enabled robust conversational agents capable of zero-shot task completions. However, applying these technologies in the biomedical field presents unique challenges. Recent initiatives like LLaVA-Med have started to adapt instruction-tuning for biomedical contexts using large datasets such as PMC-15M. Our research offers three key contributions: (i) we present a new instruct dataset enriched with medical image-text pairs from Claude3-Opus and LLaMA3 70B, (ii) we propose a novel image encoding strategy using hierarchical representations to improve fine-grained biomedical visual comprehension, and (iii) we develop the Llama3-Med model, which achieves state-of-the-art zero-shot performance on biomedical visual question answering benchmarks, with an average performance improvement of over 10% compared to previous methods. These advancements provide more accurate and reliable tools for medical professionals, bridging gaps in current multi-modal conversational assistants and promoting further innovations in medical AI.

cs.CL↗

arXiv2024

Differentially Private Fine-Tuning of Diffusion Models

The integration of Differential Privacy (DP) with diffusion models (DMs) presents a promising yet challenging frontier, particularly due to the substantial memorization capabilities of DMs that pose significant privacy risks. Differential privacy offers a rigorous framework for safeguarding individual data points during model training, with Differential Privacy Stochastic Gradient Descent (DP-SGD) being a prominent implementation. Diffusion method decomposes image generation into iterative steps, theoretically aligning well with DP's incremental noise addition. Despite the natural fit, the unique architecture of DMs necessitates tailored approaches to effectively balance privacy-utility trade-off. Recent developments in this field have highlighted the potential for generating high-quality synthetic data by pre-training on public data (i.e., ImageNet) and fine-tuning on private data, however, there is a pronounced gap in research on optimizing the trade-offs involved in DP settings, particularly concerning parameter efficiency and model scalability. Our work addresses this by proposing a parameter-efficient fine-tuning strategy optimized for private diffusion models, which minimizes the number of trainable parameters to enhance the privacy-utility trade-off. We empirically demonstrate that our method achieves state-of-the-art performance in DP synthesis, significantly surpassing previous benchmarks on widely studied datasets (e.g., with only 0.47M trainable parameters, achieving a more than 35% improvement over the previous state-of-the-art with a small privacy budget on the CelebA-64 dataset). Anonymous codes available at https://anonymous.4open.science/r/DP-LORA-F02F.

cs.CV↗

arXiv2024

Online Personalizing White-box LLMs Generation with Neural Bandits

The advent of personalized content generation by LLMs presents a novel challenge: how to efficiently adapt text to meet individual preferences without the unsustainable demand of creating a unique model for each user. This study introduces an innovative online method that employs neural bandit algorithms to dynamically optimize soft instruction embeddings based on user feedback, enhancing the personalization of open-ended text generation by white-box LLMs. Through rigorous experimentation on various tasks, we demonstrate significant performance improvements over baseline strategies. NeuralTS, in particular, leads to substantial enhancements in personalized news headline generation, achieving up to a 62.9% improvement in terms of best ROUGE scores and up to 2.76% increase in LLM-agent evaluation against the baseline.

cs.CL↗

arXiv2024

AdaptiveFL: Adaptive Heterogeneous Federated Learning for Resource-Constrained AIoT Systems

Although Federated Learning (FL) is promising to enable collaborative learning among Artificial Intelligence of Things (AIoT) devices, it suffers from the problem of low classification performance due to various heterogeneity factors (e.g., computing capacity, memory size) of devices and uncertain operating environments. To address these issues, this paper introduces an effective FL approach named AdaptiveFL based on a novel fine-grained width-wise model pruning strategy, which can generate various heterogeneous local models for heterogeneous AIoT devices. By using our proposed reinforcement learning-based device selection mechanism, AdaptiveFL can adaptively dispatch suitable heterogeneous models to corresponding AIoT devices on the fly based on their available resources for local training. Experimental results show that, compared to state-of-the-art methods, AdaptiveFL can achieve up to 16.83% inference improvements for both IID and non-IID scenarios.

cs.LG↗

arXiv2023

Observation of many-body dynamical localization

The quantum kicked rotor is a paradigmatic model system in quantum physics. As a driven quantum system, it is used to study the transition from the classical to the quantum world and to elucidate the emergence of chaos and diffusion. In contrast to its classical counterpart, it features dynamical localization, specifically Anderson localization in momentum space. The interacting many-body kicked rotor is believed to break localization, as recent experiments suggest. Here, we present evidence for many-body dynamical localization for the Lieb-Liniger version of the many-body quantum kicked rotor. After some initial evolution, the momentum distribution of interacting quantum-degenerate bosonic atoms in one-dimensional geometry, kicked hundreds of times by means of a pulsed sinusoidal potential, stops spreading. We quantify the arrested evolution by analysing the energy and the information entropy of the system as the interaction strength is tuned. In the limiting cases of vanishing and strong interactions, the first-order correlation function exhibits a very different decay behavior. Our results shed light on the boundary between the classical, chaotic world and the realm of quantum physics.

quant-ph↗

arXiv2023

A quantum register using collective excitations in a Bose-Einstein condensate

A qubit made up of an ensemble of atoms is attractive due to its resistance to atom losses, and many proposals to realize such a qubit are based on the Rydberg blockade effect. In this work, we instead consider an experimentally feasible protocol to coherently load a spin-dependent optical lattice from a spatially overlapping Bose--Einstein condensate. Identifying each lattice site as a qubit, with an empty or filled site as the qubit basis, we discuss how high-fidelity single-qubit operations, two-qubit gates between arbitrary pairs of qubits, and nondestructive measurements could be performed. In this setup, the effect of atom losses has been mitigated, the atoms never need to be removed from the ground state manifold, and separate storage and computational bases for the qubits are not required, all of which can be significant sources of decoherence in many other types of atomic qubits.

quant-ph↗

arXiv2023

Fedward: Flexible Federated Backdoor Defense Framework with Non-IID Data

Federated learning (FL) enables multiple clients to collaboratively train deep learning models while considering sensitive local datasets' privacy. However, adversaries can manipulate datasets and upload models by injecting triggers for federated backdoor attacks (FBA). Existing defense strategies against FBA consider specific and limited attacker models, and a sufficient amount of noise to be injected only mitigates rather than eliminates FBA. To address these deficiencies, we introduce a Flexible Federated Backdoor Defense Framework (Fedward) to ensure the elimination of adversarial backdoors. We decompose FBA into various attacks, and design amplified magnitude sparsification (AmGrad) and adaptive OPTICS clustering (AutoOPTICS) to address each attack. Meanwhile, Fedward uses the adaptive clipping method by regarding the number of samples in the benign group as constraints on the boundary. This ensures that Fedward can maintain the performance for the Non-IID scenario. We conduct experimental evaluations over three benchmark datasets and thoroughly compare them to state-of-the-art studies. The results demonstrate the promising defense performance from Fedward, moderately improved by 33% $\sim$ 75 in clustering defense methods, and 96.98%, 90.74%, and 89.8% for Non-IID to the utmost extent for the average FBA success rate over MNIST, FMNIST, and CIFAR10, respectively.

cs.LG↗

arXiv2023

Language Models are Few-shot Learners for Prognostic Prediction

Clinical prediction is an essential task in the healthcare industry. However, the recent success of transformers, on which large language models are built, has not been extended to this domain. In this research, we explore the use of transformers and language models in prognostic prediction for immunotherapy using real-world patients' clinical data and molecular profiles. This paper investigates the potential of transformers to improve clinical prediction compared to conventional machine learning approaches and addresses the challenge of few-shot learning in predicting rare disease areas. The study benchmarks the efficacy of baselines and language models on prognostic prediction across multiple cancer types and investigates the impact of different pretrained language models under few-shot regimes. The results demonstrate significant improvements in accuracy and highlight the potential of NLP in clinical research to improve early detection and intervention for different diseases.

cs.CL↗

arXiv2023

On the generalized Fibonacci sequence of polynomials over finite fields

In this paper, as an analogue of the integer case, we study detailedly the period and the rank of the generalized Fibonacci sequence of polynomials over a finite field modulo an arbitrary polynomial. We establish some formulas to compute them, and we also obtain some properties about the quotient of the period and the rank. We find that the polynomial case is much more complicated than the integer case.

math.NT↗

arXiv2022

Masked Image Modeling Advances 3D Medical Image Analysis

Recently, masked image modeling (MIM) has gained considerable attention due to its capacity to learn from vast amounts of unlabeled data and has been demonstrated to be effective on a wide variety of vision tasks involving natural images. Meanwhile, the potential of self-supervised learning in modeling 3D medical images is anticipated to be immense due to the high quantities of unlabeled images, and the expense and difficulty of quality labels. However, MIM's applicability to medical images remains uncertain. In this paper, we demonstrate that masked image modeling approaches can also advance 3D medical images analysis in addition to natural images. We study how masked image modeling strategies leverage performance from the viewpoints of 3D medical image segmentation as a representative downstream task: i) when compared to naive contrastive learning, masked image modeling approaches accelerate the convergence of supervised training even faster (1.40$\times$) and ultimately produce a higher dice score; ii) predicting raw voxel values with a high masking ratio and a relatively smaller patch size is non-trivial self-supervised pretext-task for medical images modeling; iii) a lightweight decoder or projection head design for reconstruction is powerful for masked image modeling on 3D medical images which speeds up training and reduce cost; iv) finally, we also investigate the effectiveness of MIM methods under different practical scenarios where different image resolutions and labeled data ratios are applied.

cs.CV↗

arXiv2022

Variational principle for optimal quantum controls in quantum metrology

We develop a variational principle to determine the quantum controls and initial state which optimizes the quantum Fisher information, the quantity characterizing the precision in quantum metrology. When the set of available controls is limited, the exact optimal initial state and the optimal controls are in general dependent on the probe time, a feature missing in the unrestricted case. Yet, for time-independent Hamiltonians with restricted controls, the problem can be approximately reduced to the unconstrained case via the Floquet engineering. In particular, we find for magnetometry with a time-independent spin chain containing three-body interactions, even when the controls are restricted to one and two-body interaction, that the Heisenberg scaling can still be approximately achieved. Our results open the door to investigate quantum metrology under a limited set of available controls, of relevance to many-body quantum metrology in realistic scenarios.

quant-ph↗