arXiv subjects

Yipeng Zhou

Publications and source records attributed to Yipeng Zhou.

At least 19 recordsLinked to original sources

arXiv2026

KREL: Automatic Medical Coding via Knowledge-Guided Reasoning over Clinical Evidence with LLMs

Automatic Medical Coding (AMC), which assigns standardized International Classification of Diseases (ICD) codes to clinical notes, is essential for medical reimbursement, quality reporting, and clinical research. Existing pre-trained language model (PLM)-based methods typically formulate AMC as an extreme multi-label classification problem over a predefined code set, while recent large language model (LLM)-based approaches instead frame it as generation or multi-step reasoning. However, key challenges remain, including the extreme length of clinical notes that hinders effective interpretation, the vast ICD label space, and complex coding rules that are not explicitly captured by LLMs. In this work, we propose Knowledge-Guided Reasoning over Clinical Evidence with LLMs (KREL), a framework that leverages LLMs for clinical text understanding and reasoning while integrating external ICD coding guidelines as structured knowledge. This design enables tight coupling between domain knowledge and LLM reasoning, reducing hallucinations and improving compliance with coding standards. Experiments on benchmark datasets show that KREL consistently outperforms strong PLM-based and state-of-the-art LLM-based baselines.

cs.CL↗

arXiv2026

Adapter-Augmented Bandits for Online Multi-Constrained Multi-Modal Inference Scheduling

Multi-modal large language model (MLLM) inference scheduling enables strong response quality under practical and heterogeneous budgets, beyond what a homogeneous single-backend setting can offer. Yet online MLLM task scheduling is nontrivial, as requests vary sharply in modality composition and latent reasoning difficulty, while execution backends incur distinct, time-varying costs due to system jitter and network variation. These coupled uncertainties pose two core challenges: deriving semantically faithful yet scheduling-relevant multi-modal task representations, and making low-overhead online decisions over irreversible multi-dimensional budgets. Accordingly, we propose \emph{M-CMAB} (\underline{M}ulti-modal \underline{M}ulti-constraint \underline{C}ontextual \underline{M}ulti-\underline{A}rmed \underline{B}andit), a multi-adapter-enhanced MLLM inference scheduling framework with three components: (i) a CLS-attentive, frozen-backbone \emph{Predictor} that extracts compact task representations and updates only lightweight adapters for action-specific estimation; (ii) a primal-dual \emph{Constrainer} that maintains online Lagrange multipliers to enforce long-horizon constraints via per-round objectives; and (iii) a two-phase \emph{Scheduler} that balances exploration and exploitation under irreversible budgets. We establish a regret guarantee under multi-dimensional knapsack constraints. On a composite multimodal benchmark with heterogeneous backends, \emph{M-CMAB} consistently outperforms state-of-the-art baselines across budget regimes, achieving up to 14.18% higher reward and closely tracking an oracle-aided upper bound. Codes are available at https://anonymous.4open.science/r/M2CMAB/.

cs.LG↗

arXiv2026

Federated Learning at the Forefront of Fairness: A Multifaceted Perspective

Fairness in Federated Learning (FL) is emerging as a critical factor driven by heterogeneous clients' constraints and balanced model performance across various scenarios. In this survey, we delineate a comprehensive classification of the state-of-the-art fairness-aware approaches from a multifaceted perspective, i.e., model performance-oriented and capability-oriented. Moreover, we provide a framework to categorize and address various fairness concerns and associated technical aspects, examining their effectiveness in balancing equity and performance within FL frameworks. We further examine several significant evaluation metrics leveraged to measure fairness quantitatively. Finally, we explore exciting open research directions and propose prospective solutions that could drive future advancements in this important area, laying a solid foundation for researchers working toward fairness in FL.

cs.LG↗

arXiv2025

Differentially Private Federated Low Rank Adaptation Beyond Fixed-Matrix

Large language models (LLMs) typically require fine-tuning for domain-specific tasks, and LoRA offers a computationally efficient approach by training low-rank adapters. LoRA is also communication-efficient for federated LLMs when multiple users collaboratively fine-tune a global LLM model without sharing their proprietary raw data. However, even the transmission of local adapters between a server and clients risks serious privacy leakage. Applying differential privacy (DP) to federated LoRA encounters a dilemma: adding noise to both adapters amplifies synthetic noise on the model, while fixing one adapter impairs the learnability of fine-tuning. In this paper, we propose FedASK (Differentially Private Federated Low Rank Adaptation with Double Sketching) , a novel federated LoRA framework to enable effective updating of both low-rank adapters with robust differential privacy. Inspired by randomized SVD, our key idea is a two-stage sketching pipeline. This pipeline first aggregates carefully sketched, privacy-preserving local updates, and then reconstructs the global matrices on the server to facilitate effective updating of both adapters. We theoretically prove FedASK's differential privacy guarantee and its exact aggregation property. Comprehensive experiments demonstrate that FedASK consistently outperforms baseline methods across a variety of privacy settings and data distributions.

cs.CR↗

arXiv2025

Siamese Foundation Models for Crystal Structure Prediction

Predicting crystal structures from chemical compositions is a fundamental challenge in materials discovery, complicated by complex 3D geometries that distinguish it from fields like protein folding. Here, we present Diffusion-based Crystal Omni (DAO), a pretrain-finetune framework for crystal structure prediction integrating two Siamese foundation models: a structure generator and an energy predictor. The generator is pretrained via a two-stage pipeline on a vast dataset of stable and unstable structures, leveraging the predictor to relax unstable configurations and guide the generative sampling. Across two well-known benchmarks, pretraining significantly enhances performance across multiple backbone architectures. Ablation studies confirm that the synergy between the generator and predictor mutually benefits both components. We further validate DAO on three real-world superconductors ($\text{Cr}_6\text{Os}_2$, $\text{Zr}_{16}\text{Rh}_8\text{O}_4$, and $\text{Zr}_{16}\text{Pd}_8\text{O}_4$) typically inaccessible to conventional computation. For $\text{Cr}_6\text{Os}_2$, DAO achieves a 100\% match rate with experimental references and an atomic-position error of 0.0012 under 20-shot generation, performing over 2000$\times$ faster per iteration than DFT-based structure predictors. These compelling results collectively highlight the potential of our approach for advancing materials science research.

cond-mat.mtrl-sci↗

arXiv2025

ExClique: An Express Consensus Algorithm for High-Speed Transaction Process in Blockchains

Proof of Authority (PoA) plays a pivotal role in blockchains for reaching consensus. Clique, which selects consensus nodes to generate blocks with a pre-determined order, is the most popular implementation of PoA due to its low communication overhead and energy consumption. However, our study unveils that the speed to process transactions by Clique is severely restricted by 1) the long communication delay of full blocks (each containing a certain number of transactions) between consensus nodes; and 2) occurrences of no-turn blocks, generated by no-turn nodes if an in-turn block generation fails. Consequently, Clique struggles to support distributed applications requiring a high transaction processing speed, e.g., online gaming. To overcome this deficiency, we propose an Express Clique (ExClique) algorithm by improving Clique from two perspectives: compacting blocks for broadcasting to shorten communication delay and prohibiting the occurrences of no-turn blocks. For performance evaluation, we implement ExClique by modifying Geth of Ethereum, the software implementing Clique, and deploy a permissioned blockchain network by using container technology. The experimental results show that ExClique achieves a substantial enhancement in transactions per second (TPS). Specifically, it boosts TPS by 2.25X in a typical network with 21 consensus nodes and an impressive 7.01X in a large-scale network with 101 consensus nodes when compared to Clique.

cs.DC↗

arXiv2025

MiniMax-01: Scaling Foundation Models with Lightning Attention

We introduce MiniMax-01 series, including MiniMax-Text-01 and MiniMax-VL-01, which are comparable to top-tier models while offering superior capabilities in processing longer contexts. The core lies in lightning attention and its efficient scaling. To maximize computational capacity, we integrate it with Mixture of Experts (MoE), creating a model with 32 experts and 456 billion total parameters, of which 45.9 billion are activated for each token. We develop an optimized parallel strategy and highly efficient computation-communication overlap techniques for MoE and lightning attention. This approach enables us to conduct efficient training and inference on models with hundreds of billions of parameters across contexts spanning millions of tokens. The context window of MiniMax-Text-01 can reach up to 1 million tokens during training and extrapolate to 4 million tokens during inference at an affordable cost. Our vision-language model, MiniMax-VL-01 is built through continued training with 512 billion vision-language tokens. Experiments on both standard and in-house benchmarks show that our models match the performance of state-of-the-art models like GPT-4o and Claude-3.5-Sonnet while offering 20-32 times longer context window. We publicly release MiniMax-01 at https://github.com/MiniMax-AI.

cs.CL↗

arXiv2024

Metal-strong Inflows at the Outer-galactic-scale of a Quasar

We present an analysis of the absorption-line system in the Very Large Telescope/Ultraviolet and Visual Echelle Spectrograph spectrum at a redshift of $z_{\rm a}={3.1448}$ associated with the quasar SDSS J122040.23+092326.96, whose systematic redshift is $z_{\rm e}=3.1380\pm0.0007$, measured from the ${\rm H}\beta$+[O III] emission lines in our newly acquired NIR P200/TripleSpec data. This absorbing system, detected in numerous absorption lines including the N V, N III, C IV, C III, Si IV, Si III, and H I Lyman series, can be resolved into seven kinematic components with red-shifted velocities ranging from 200 to $900\,\rm km\,s^{-1}$. The high-ionization N V doublet detected and the rather narrow Lyman series measured ($b\approx14\,\rm km\,s^{-1}$) suggest that the absorption gas is photo ionized, possibly by the quasar. A low density is inferred by the fact that N III $\lambda989.80$ is significantly detected while N III* $\lambda991.51$ (${\rm log}\,n_{\rm c}=3.3\,\rm cm^{-3}$) is undetectably weak. A firm lower limit of a solar value to the abundance of the gas can be set based on the measurements of Si IV and H I column densities, as first proposed by F. Hamann. Detailed photoionization simulations indicate that $T1$, and possibly the absorber as a whole, has metallicities of $Z\sim1.5-6.0\,Z\rm\,sun$, and is located at $\sim15\,\rm kpc$ from the quasar nucleus. The metal-strong absorption inflows at the outskirt of the quasar host galaxy is most likely originated in situ and were driven by stellar processes, such as stellar winds and/or supernova explosions. Such a relatively rare system may hold important clues to understanding the baryonic cycling of galaxies, and more cases could be picked out using relatively strong Si IV and weak Lyman absorption lines.

astro-ph.GA↗

arXiv2024

BGTplanner: Maximizing Training Accuracy for Differentially Private Federated Recommenders via Strategic Privacy Budget Allocation

To mitigate the rising concern about privacy leakage, the federated recommender (FR) paradigm emerges, in which decentralized clients co-train the recommendation model without exposing their raw user-item rating data. The differentially private federated recommender (DPFR) further enhances FR by injecting differentially private (DP) noises into clients. Yet, current DPFRs, suffering from noise distortion, cannot achieve satisfactory accuracy. Various efforts have been dedicated to improving DPFRs by adaptively allocating the privacy budget over the learning process. However, due to the intricate relation between privacy budget allocation and model accuracy, existing works are still far from maximizing DPFR accuracy. To address this challenge, we develop BGTplanner (Budget Planner) to strategically allocate the privacy budget for each round of DPFR training, improving overall training performance. Specifically, we leverage the Gaussian process regression and historical information to predict the change in recommendation accuracy with a certain allocated privacy budget. Additionally, Contextual Multi-Armed Bandit (CMAB) is harnessed to make privacy budget allocation decisions by reconciling the current improvement and long-term privacy constraints. Our extensive experimental results on real datasets demonstrate that \emph{BGTplanner} achieves an average improvement of 6.76\% in training performance compared to state-of-the-art baselines.

cs.LG↗

arXiv2024

A Communication and Computation Efficient Fully First-order Method for Decentralized Bilevel Optimization

Bilevel optimization, crucial for hyperparameter tuning, meta-learning and reinforcement learning, remains less explored in the decentralized learning paradigm, such as decentralized federated learning (DFL). Typically, decentralized bilevel methods rely on both gradients and Hessian matrices to approximate hypergradients of upper-level models. However, acquiring and sharing the second-order oracle is compute and communication intensive. % and sharing this information incurs heavy communication overhead. To overcome these challenges, this paper introduces a fully first-order decentralized method for decentralized Bilevel optimization, $\text{C}^2$DFB which is both compute- and communicate-efficient. In $\text{C}^2$DFB, each learning node optimizes a min-min-max problem to approximate hypergradient by exclusively using gradients information. To reduce the traffic load at the inner-loop of solving the lower-level problem, $\text{C}^2$DFB incorporates a lightweight communication protocol for efficiently transmitting compressed residuals of local parameters. % during the inner loops. Rigorous theoretical analysis ensures its convergence % of the algorithm, indicating a first-order oracle calls of $\tilde{\mathcal{O}}(\epsilon^{-4})$. Experiments on hyperparameter tuning and hyper-representation tasks validate the superiority of $\text{C}^2$DFB across various typologies and heterogeneous data distributions.

cs.LG↗

arXiv2024

Snapshot: Towards Application-centered Models for Pedestrian Trajectory Prediction in Urban Traffic Environments

This paper explores pedestrian trajectory prediction in urban traffic while focusing on both model accuracy and real-world applicability. While promising approaches exist, they often revolve around pedestrian datasets excluding traffic-related information, or resemble architectures that are either not real-time capable or robust. To address these limitations, we first introduce a dedicated benchmark based on Argoverse 2, specifically targeting pedestrians in traffic environments. Following this, we present Snapshot, a modular, feed-forward neural network that outperforms the current state of the art, reducing the Average Displacement Error (ADE) by 8.8% while utilizing significantly less information. Despite its agent-centric encoding scheme, Snapshot demonstrates scalability, real-time performance, and robustness to varying motion histories. Moreover, by integrating Snapshot into a modular autonomous driving software stack, we showcase its real-world applicability.

cs.CV↗

arXiv2024

PPVF: An Efficient Privacy-Preserving Online Video Fetching Framework with Correlated Differential Privacy

Online video streaming has evolved into an integral component of the contemporary Internet landscape. Yet, the disclosure of user requests presents formidable privacy challenges. As users stream their preferred online videos, their requests are automatically seized by video content providers, potentially leaking users' privacy. Unfortunately, current protection methods are not well-suited to preserving user request privacy from content providers while maintaining high-quality online video services. To tackle this challenge, we introduce a novel Privacy-Preserving Video Fetching (PPVF) framework, which utilizes trusted edge devices to pre-fetch and cache videos, ensuring the privacy of users' requests while optimizing the efficiency of edge caching. More specifically, we design PPVF with three core components: (1) \textit{Online privacy budget scheduler}, which employs a theoretically guaranteed online algorithm to select non-requested videos as candidates with assigned privacy budgets. Alternative videos are chosen by an online algorithm that is theoretically guaranteed to consider both video utilities and available privacy budgets. (2) \textit{Noisy video request generator}, which generates redundant video requests (in addition to original ones) utilizing correlated differential privacy to obfuscate request privacy. (3) \textit{Online video utility predictor}, which leverages federated learning to collaboratively evaluate video utility in an online fashion, aiding in video selection in (1) and noise generation in (2). Finally, we conduct extensive experiments using real-world video request traces from Tencent Video. The results demonstrate that PPVF effectively safeguards user request privacy while upholding high video caching performance.

cs.MM↗

arXiv2024

Mitigating Noise Detriment in Differentially Private Federated Learning with Model Pre-training

Differentially Private Federated Learning (DPFL) strengthens privacy protection by perturbing model gradients with noise, though at the cost of reduced accuracy. Although prior empirical studies indicate that initializing from pre-trained rather than random parameters can alleviate noise disturbance, the problem of optimally fine-tuning pre-trained models in DPFL remains unaddressed. In this paper, we propose Pretrain-DPFL, a framework that systematically evaluates three most representative fine-tuning strategies: full-tuning (FT), head-tuning (HT), and unified-tuning(UT) combining HT followed by FT. Through convergence analysis under smooth non-convex loss, we establish theoretical conditions for identifying the optimal fine-tuning strategy in Pretrain-DPFL, thereby maximizing the benefits of pre-trained models in mitigating noise disturbance. Extensive experiments across multiple datasets demonstrate Pretrain-DPFL's superiority, achieving $25.22\%$ higher accuracy than scratch training and outperforming the second-best baseline by $8.19\%$, significantly improving the privacy-utility trade-off in DPFL.

cs.LG↗

arXiv2024

The Power of Bias: Optimizing Client Selection in Federated Learning with Heterogeneous Differential Privacy

To preserve the data privacy, the federated learning (FL) paradigm emerges in which clients only expose model gradients rather than original data for conducting model training. To enhance the protection of model gradients in FL, differentially private federated learning (DPFL) is proposed which incorporates differentially private (DP) noises to obfuscate gradients before they are exposed. Yet, an essential but largely overlooked problem in DPFL is the heterogeneity of clients' privacy requirement, which can vary significantly between clients and extremely complicates the client selection problem in DPFL. In other words, both the data quality and the influence of DP noises should be taken into account when selecting clients. To address this problem, we conduct convergence analysis of DPFL under heterogeneous privacy, a generic client selection strategy, popular DP mechanisms and convex loss. Based on convergence analysis, we formulate the client selection problem to minimize the value of loss function in DPFL with heterogeneous privacy, which is a convex optimization problem and can be solved efficiently. Accordingly, we propose the DPFL-BCS (biased client selection) algorithm. The extensive experiment results with real datasets under both convex and non-convex loss functions indicate that DPFL-BCS can remarkably improve model utility compared with the SOTA baselines.

cs.LG↗

arXiv2024

DEthna: Accurate Ethereum Network Topology Discovery with Marked Transactions

In Ethereum, the ledger exchanges messages along an underlying Peer-to-Peer (P2P) network to reach consistency. Understanding the underlying network topology of Ethereum is crucial for network optimization, security and scalability. However, the accurate discovery of Ethereum network topology is non-trivial due to its deliberately designed security mechanism. Consequently, existing measuring schemes cannot accurately infer the Ethereum network topology with a low cost. To address this challenge, we propose the Distributed Ethereum Network Analyzer (DEthna) tool, which can accurately and efficiently measure the Ethereum network topology. In DEthna, a novel parallel measurement model is proposed that can generate marked transactions to infer link connections based on the transaction replacement and propagation mechanism in Ethereum. Moreover, a workload offloading scheme is designed so that DEthna can be deployed on multiple distributed probing nodes so as to measure a large-scale Ethereum network at a low cost. We run DEthna on Goerli (the most popular Ethereum test network) to evaluate its capability in discovering network topology. The experimental results demonstrate that DEthna significantly outperforms the state-of-the-art baselines. Based on DEthna, we further analyze characteristics of the Ethereum network revealing that there exist more than 50% low-degree Ethereum nodes that weaken the network robustness.

cs.NI↗

arXiv2024

A Survey on Privacy-Preserving Caching at Network Edge: Classification, Solutions, and Challenges

Caching content at the edge network is a popular and effective technique widely deployed to alleviate the burden of network backhaul, shorten service delay and improve service quality. However, there has been some controversy over privacy violations in caching content at the edge network. On the one hand, the multi-access open edge network provides an ideal entrance or interface for external attackers to obtain private data from edge caches by extracting sensitive information. On the other hand, privacy can be infringed on by curious edge caching providers through caching trace analysis targeting the achievement of better caching performance or higher profits. Therefore, an in-depth understanding of privacy issues in edge caching networks is vital and indispensable for creating a privacy-preserving caching service at the edge network. In this article, we are among the first to fill this gap by examining privacy-preserving techniques for caching content at the edge network. Firstly, we provide an introduction to the background of privacy-preserving edge caching (PPEC). Next, we summarize the key privacy issues and present a taxonomy for caching at the edge network from the perspective of private information. Additionally, we conduct a retrospective review of the state-of-the-art countermeasures against privacy leakage from content caching at the edge network. Finally, we conclude the survey and envision challenges for future research.

cs.NI↗

arXiv2024

Fed-CVLC: Compressing Federated Learning Communications with Variable-Length Codes

In Federated Learning (FL) paradigm, a parameter server (PS) concurrently communicates with distributed participating clients for model collection, update aggregation, and model distribution over multiple rounds, without touching private data owned by individual clients. FL is appealing in preserving data privacy; yet the communication between the PS and scattered clients can be a severe bottleneck. Model compression algorithms, such as quantization and sparsification, have been suggested but they generally assume a fixed code length, which does not reflect the heterogeneity and variability of model updates. In this paper, through both analysis and experiments, we show strong evidences that variable-length is beneficial for compression in FL. We accordingly present Fed-CVLC (Federated Learning Compression with Variable-Length Codes), which fine-tunes the code length in response of the dynamics of model updates. We develop optimal tuning strategy that minimizes the loss function (equivalent to maximizing the model utility) subject to the budget for communication. We further demonstrate that Fed-CVLC is indeed a general compression design that bridges quantization and sparsification, with greater flexibility. Extensive experiments have been conducted with public datasets to demonstrate that Fed-CVLC remarkably outperforms state-of-the-art baselines, improving model utility by 1.50%-5.44%, or shrinking communication traffic by 16.67%-41.61%.

cs.LG↗

arXiv2024

Expediting In-Network Federated Learning by Voting-Based Consensus Model Compression

Recently, federated learning (FL) has gained momentum because of its capability in preserving data privacy. To conduct model training by FL, multiple clients exchange model updates with a parameter server via Internet. To accelerate the communication speed, it has been explored to deploy a programmable switch (PS) in lieu of the parameter server to coordinate clients. The challenge to deploy the PS in FL lies in its scarce memory space, prohibiting running memory consuming aggregation algorithms on the PS. To overcome this challenge, we propose Federated Learning in-network Aggregation with Compression (FediAC) algorithm, consisting of two phases: client voting and model aggregating. In the former phase, clients report their significant model update indices to the PS to estimate global significant model updates. In the latter phase, clients upload global significant model updates to the PS for aggregation. FediAC consumes much less memory space and communication traffic than existing works because the first phase can guarantee consensus compression across clients. The PS easily aligns model update indices to swiftly complete aggregation in the second phase. Finally, we conduct extensive experiments by using public datasets to demonstrate that FediAC remarkably surpasses the state-of-the-art baselines in terms of model accuracy and communication traffic.

cs.LG↗