Dong-Jun Han

Wenzhi Fang, Dong-Jun Han, Evan Chen et al.

While traditional federated learning (FL) typically focuses on a star topology where clients are directly connected to a central server, real-world distributed systems often exhibit hierarchical architectures. Hierarchical FL (HFL) has emerged as a promising solution to bridge this gap, leveraging aggregation points at multiple levels of the system. However, existing algorithms for HFL encounter challenges in dealing with multi-timescale model drift, i.e., model drift occurring across hierarchical levels of data heterogeneity. In this paper, we propose a multi-timescale gradient correction (MTGC) methodology to resolve this issue. Our key idea is to introduce distinct control variables to (i) correct the client gradient towards the group gradient, i.e., to reduce client model drift caused by local updates based on individual datasets, and (ii) correct the group gradient towards the global gradient, i.e., to reduce group model drift caused by FL over clients within the group. We analytically characterize the convergence behavior of MTGC under general non-convex settings, overcoming challenges associated with couplings between correction terms. We show that our convergence bound is immune to the extent of data heterogeneity, confirming the stability of the proposed algorithm against multi-level non-i.i.d. data. Through extensive experiments on various datasets and models, we validate the effectiveness of MTGC in diverse HFL settings. The code for this project is available at \href{https://github.com/wenzhifang/MTGC}{https://github.com/wenzhifang/MTGC}.

Dong-Jun Han, Do-Yeon Kim, Minseok Choi et al.

A fundamental challenge to providing edge-AI services is the need for a machine learning (ML) model that achieves personalization (i.e., to individual clients) and generalization (i.e., to unseen data) properties concurrently. Existing techniques in federated learning (FL) have encountered a steep tradeoff between these objectives and impose large computational requirements on edge devices during training and inference. In this paper, we propose SplitGP, a new split learning solution that can simultaneously capture generalization and personalization capabilities for efficient inference across resource-constrained clients (e.g., mobile/IoT devices). Our key idea is to split the full ML model into client-side and server-side components, and impose different roles to them: the client-side model is trained to have strong personalization capability optimized to each client's main task, while the server-side model is trained to have strong generalization capability for handling all clients' out-of-distribution tasks. We analytically characterize the convergence behavior of SplitGP, revealing that all client models approach stationary points asymptotically. Further, we analyze the inference time in SplitGP and provide bounds for determining model split ratios. Experimental results show that SplitGP outperforms existing baselines by wide margins in inference time and test accuracy for varying amounts of out-of-distribution samples.

Jungwuk Park, Dong-Jun Han, Soyeong Kim et al.

In domain generalization (DG), the target domain is unknown when the model is being trained, and the trained model should successfully work on an arbitrary (and possibly unseen) target domain during inference. This is a difficult problem, and despite active studies in recent years, it remains a great challenge. In this paper, we take a simple yet effective approach to tackle this issue. We propose test-time style shifting, which shifts the style of the test sample (that has a large style gap with the source domains) to the nearest source domain that the model is already familiar with, before making the prediction. This strategy enables the model to handle any target domains with arbitrary style statistics, without additional model update at test-time. Additionally, we propose style balancing, which provides a great platform for maximizing the advantage of test-time style shifting by handling the DG-specific imbalance issues. The proposed ideas are easy to implement and successfully work in conjunction with various other DG schemes. Experimental results on different datasets show the effectiveness of our methods.

Jungwuk Park, Dong-Jun Han, Jinho Kim et al.

Traditional federated learning (FL) algorithms operate under the assumption that the data distributions at training (source domains) and testing (target domain) are the same. The fact that domain shifts often occur in practice necessitates equipping FL methods with a domain generalization (DG) capability. However, existing DG algorithms face fundamental challenges in FL setups due to the lack of samples/domains in each client's local dataset. In this paper, we propose StableFDG, a style and attention based learning strategy for accomplishing federated domain generalization, introducing two key contributions. The first is style-based learning, which enables each client to explore novel styles beyond the original source domains in its local dataset, improving domain diversity based on the proposed style sharing, shifting, and exploration strategies. Our second contribution is an attention-based feature highlighter, which captures the similarities between the features of data samples in the same class, and emphasizes the important/common characteristics to better learn the domain-invariant characteristics of each class in data-poor FL scenarios. Experimental results show that StableFDG outperforms existing baselines on various DG benchmark datasets, demonstrating its efficacy.

Yun-Wei Chu, Dong-Jun Han, Seyyedali Hosseinalipour et al.

Over the past several years, various federated learning (FL) methodologies have been developed to improve model accuracy, a primary performance metric in machine learning. However, to utilize FL in practical decision-making scenarios, beyond considering accuracy, the trained model must also have a reliable confidence in each of its predictions, an aspect that has been largely overlooked in existing FL research. Motivated by this gap, we propose Non-Uniform Calibration for Federated Learning (NUCFL), a generic framework that integrates FL with the concept of model calibration. The inherent data heterogeneity in FL environments makes model calibration particularly difficult, as it must ensure reliability across diverse data distributions and client conditions. Our NUCFL addresses this challenge by dynamically adjusting the model calibration objectives based on statistical relationships between each client's local model and the global model in FL. In particular, NUCFL assesses the similarity between local and global model relationships, and controls the penalty term for the calibration loss during client-side local training. By doing so, NUCFL effectively aligns calibration needs for the global model in heterogeneous FL settings while not sacrificing accuracy. Extensive experiments show that NUCFL offers flexibility and effectiveness across various FL algorithms, enhancing accuracy as well as model calibration.

Seokil Ham, Jungwuk Park, Dong-Jun Han et al.

While multi-exit neural networks are regarded as a promising solution for making efficient inference via early exits, combating adversarial attacks remains a challenging problem. In multi-exit networks, due to the high dependency among different submodels, an adversarial example targeting a specific exit not only degrades the performance of the target exit but also reduces the performance of all other exits concurrently. This makes multi-exit networks highly vulnerable to simple adversarial attacks. In this paper, we propose NEO-KD, a knowledge-distillation-based adversarial training strategy that tackles this fundamental challenge based on two key contributions. NEO-KD first resorts to neighbor knowledge distillation to guide the output of the adversarial examples to tend to the ensemble outputs of neighbor exits of clean data. NEO-KD also employs exit-wise orthogonal knowledge distillation for reducing adversarial transferability across different submodels. The result is a significantly improved robustness against adversarial attacks. Experimental results on various datasets/models show that our method achieves the best adversarial accuracy with reduced computation budgets, compared to the baselines relying on existing adversarial training or knowledge distillation techniques for multi-exit networks.

Seohyun Lee, Wenzhi Fang, Dong-Jun Han et al.

Recent works have advanced feedback-based learning systems, whereby a foundation model is able to intake incoming feedback (e.g., a user) to self-improve, creating a self-loop system of training. However, existing works are limited in needing to consider an offline setup to allow for such feedback-based methods, and are further limited in the need of requiring privileged ground-truth contexts for training. Moreover, there is limited consideration of federated learning (FL), which is particularly well-suited for incorporating external feedback across large networks of end users, for example, but requires methods to be efficient for training on resource-constrained edge devices. Therefore, we introduce SPEAR (Self-Play Enhancement via Advantage-Weighted Refinement), an efficient online learning algorithm for federated LLM fine-tuning. SPEAR utilizes a feedback-guided self-play loop to construct naturally contrastive pairs per prompt which are utilized to be trained on (i) standard maximum likelihood on correct completions and (ii) confidence-weighted unlikelihood on tail tokens of incorrect completions. Without the need of expensive group generations and ground-truth contexts for training (i.e., only partial, non-answer feedback), in contrast with existing works, SPEAR can be trained both online and in a resource-efficient manner. We validate SPEAR across various benchmark datasets, demonstrating its superior performance in comparison to state-of-the-art baselines. The implementation code is publicly available at https://github.com/lee3296/SPEAR.

Liangqi Yuan, Dong-Jun Han, Vishnu Pandi Chellapandi et al.

Multimodal federated learning (FL) aims to enrich model training in FL settings where devices are collecting measurements across multiple modalities (e.g., sensors measuring pressure, motion, and other types of data). However, key challenges to multimodal FL remain unaddressed, particularly in heterogeneous network settings: (i) the set of modalities collected by each device will be diverse, and (ii) communication limitations prevent devices from uploading all their locally trained modality models to the server. In this paper, we propose Federated Multimodal Fusion learning with Selective modality communication (FedMFS), a new multimodal fusion FL methodology that can tackle the above mentioned challenges. The key idea is the introduction of a modality selection criterion for each device, which weighs (i) the impact of the modality, gauged by Shapley value analysis, against (ii) the modality model size as a gauge for communication overhead. This enables FedMFS to flexibly balance performance against communication costs, depending on resource constraints and application requirements. Experiments on the real-world ActionSense dataset demonstrate the ability of FedMFS to achieve comparable accuracy to several baselines while reducing the communication overhead by over 4x.

Wenzhi Fang, Dong-Jun Han, Christopher G. Brinton

Hierarchical federated learning (HFL) has demonstrated promising scalability advantages over the traditional "star-topology" architecture-based federated learning (FL). However, HFL still imposes significant computation, communication, and storage burdens on the edge, especially when training a large-scale model over resource-constrained wireless devices. In this paper, we propose hierarchical independent submodel training (HIST), a new FL methodology that aims to address these issues in hierarchical cloud-edge-client networks. The key idea behind HIST is to divide the global model into disjoint partitions (or submodels) per round so that each group of clients (i.e., cells) is responsible for training only one partition of the model. We characterize the convergence behavior of HIST under mild assumptions, showing the impacts of several key attributes (e.g., submodel sizes, number of cells, edge and global aggregation frequencies) on the rate and stationarity gap. Building upon the theoretical results, we propose a submodel partitioning strategy to minimize the training latency depending on network resource availability and a target learning performance guarantee. We then demonstrate how HIST can be augmented with over-the-air computation (AirComp) to further enhance the efficiency of the model aggregation over the edge cells. Through numerical evaluations, we verify that HIST is able to save training time and communication costs by wide margins while achieving comparable accuracy as conventional HFL. Moreover, our experiments demonstrate that AirComp-assisted HIST provides further improvements in training latency.

Hyeonjin Kim, Hangyeol Jung, Heechan Yun et al.

Unlearning specific concepts in text-to-image diffusion models has become increasingly important for preventing undesirable content generation. Among prior approaches, sparse autoencoder (SAE)-based methods have attracted attention due to their ability to suppress target concepts through lightweight manipulation of latent features, without modifying model parameters. However, SAEs trained with sparse reconstruction objectives do not explicitly enforce concept-wise separation, resulting in shared latent features across concepts. To address this, we propose SAEParate, which organizes latent representations into concept-specific clusters via a concept-aware contrastive objective, enabling more precise concept suppression while reducing unintended interference during unlearning. In addition, we enhance the encoder with a GeLU-based nonlinear transformation to increase its expressive capacity under this separation objective, enabling a more discriminative and disentangled latent space. Experiments on UnlearnCanvas demonstrate state-of-the-art performance, with particularly strong gains in joint style-object unlearning, a challenging setting where existing methods suffer from severe interference between target and non-target concepts.

Seohyun Lee, Wenzhi Fang, Dong-Jun Han et al.

Federated Learning (FL), despite demonstrating impressive capabilities in the training of multiple models in a decentralized manner, has been shown to produce a final model not necessarily well-suited to the needs of each client. While extensive work has been conducted on how to create tailored personalized models, called Personalized Federated Learning (PFL), less attention has been given to personalization via fine-tuning of foundation models with multi-task and multi-modal properties. Moreover, there exists a lack of understanding in the literature on how to fine-tune and personalize such models in a setting that is heterogeneous across clients not only in data, but also in tasks and modalities. To address this gap in the literature, we propose TAP (Two-Stage Adaptive Personalization), which (i) leverages mismatched model architectures between the clients and server to selectively conduct replacement operations when it benefits a client's local tasks and (ii) engages in post-FL knowledge distillation for capturing beneficial general knowledge without compromising personalization. We also introduce the first convergence analysis of the server model under its modality-task pair architecture, and demonstrate that as the number of modality-task pairs increases, its ability to cater to all tasks suffers. Through extensive experiments, we demonstrate the effectiveness of our proposed algorithm across a variety of datasets and tasks in comparison to a multitude of baselines. Implementation code is publicly available at https://github.com/lee3296/TAP.

Wenzhi Fang, Liangqi Yuan, Guangchen Lan et al.

Multi-agent large language model (LLM) systems often rely on a controller to coordinate a pool of heterogeneous models, yet existing controllers are typically limited to one-shot routing: they select a model once and return its output directly. Such routing-only designs provide no mechanism to critique intermediate drafts or support iterative refinement. To address this limitation, we propose a critique-and-routing controller that casts multi-agent coordination as a sequential decision problem. At each turn, the controller evaluates the current draft, decides whether to stop or continue, and, if needed, selects the next agent for further refinement. We formulate this process as a finite-horizon Markov Decision Process (MDP) with explicit agent-utilization constraints, design a composite reward for controller decisions across turns, and optimize the controller via policy gradients under a Lagrangian-relaxed objective. Extensive experiments across multiple heterogeneous multi-agent systems and seven reasoning benchmarks show that our method consistently outperforms state-of-the-art baselines and substantially narrows the gap to the strongest agent, while using it for fewer than 25% of total calls.

Guangchen Lan, Dong-Jun Han, Abolfazl Hashemi et al.

To improve the efficiency of reinforcement learning (RL), we propose a novel asynchronous federated reinforcement learning (FedRL) framework termed AFedPG, which constructs a global model through collaboration among $N$ agents using policy gradient (PG) updates. To address the challenge of lagged policies in asynchronous settings, we design a delay-adaptive lookahead technique \textit{specifically for FedRL} that can effectively handle heterogeneous arrival times of policy gradients. We analyze the theoretical global convergence bound of AFedPG, and characterize the advantage of the proposed algorithm in terms of both the sample complexity and time complexity. Specifically, our AFedPG method achieves $O(\frac{ε^{-2.5}}{N})$ sample complexity for global convergence at each agent on average. Compared to the single agent setting with $O(ε^{-2.5})$ sample complexity, it enjoys a linear speedup with respect to the number of agents. Moreover, compared to synchronous FedPG, AFedPG improves the time complexity from $O(\frac{t_{\max}}{N})$ to $O({\sum_{i=1}^{N} \frac{1}{t_{i}}})^{-1}$, where $t_{i}$ denotes the time consumption in each iteration at agent $i$, and $t_{\max}$ is the largest one. The latter complexity $O({\sum_{i=1}^{N} \frac{1}{t_{i}}})^{-1}$ is always smaller than the former one, and this improvement becomes significant in large-scale federated settings with heterogeneous computing powers ($t_{\max}\gg t_{\min}$). Finally, we empirically verify the improved performance of AFedPG in four widely used MuJoCo environments with varying numbers of agents. We also demonstrate the advantages of AFedPG in various computing heterogeneity scenarios.

Liangqi Yuan, Dong-Jun Han, Su Wang et al.

Multimodal federated learning (FL) aims to enrich model training in FL settings where clients are collecting measurements across multiple modalities. However, key challenges to multimodal FL remain unaddressed, particularly in heterogeneous network settings where: (i) the set of modalities collected by each client will be diverse, and (ii) communication limitations prevent clients from uploading all their locally trained modality models to the server. In this paper, we propose multimodal Federated learning with joint Modality and Client selection (mmFedMC), a new FL methodology that can tackle the above-mentioned challenges in multimodal settings. The joint selection algorithm incorporates two main components: (a) A modality selection methodology for each client, which weighs (i) the impact of the modality, gauged by Shapley value analysis, (ii) the modality model size as a gauge of communication overhead, against (iii) the frequency of modality model updates, denoted recency, to enhance generalizability. (b) A client selection strategy for the server based on the local loss of modality model at each client. Experiments on five real-world datasets demonstrate the ability of mmFedMC to achieve comparable accuracy to several baselines while reducing the communication overhead by over 20x. A demo video of our methodology is available at https://liangqiy.com/mmfedmc/.

Liangqi Yuan, Dong-Jun Han, Shiqiang Wang et al.

Compared to traditional machine learning models, recent large language models (LLMs) can exhibit multi-task-solving capabilities through multiple dialogues and multi-modal data sources. These unique characteristics of LLMs, together with their large model size, make their deployment more challenging. Specifically, (i) deploying LLMs on local devices faces computational, memory, and energy resource issues, while (ii) deploying them in the cloud cannot guarantee real-time service and incurs communication/usage costs. In this paper, we design TMO, a local-cloud LLM inference system with Three-M Offloading: Multi-modal, Multi-task, and Multi-dialogue. TMO incorporates (i) a lightweight local LLM that can process simple tasks at high speed and (ii) a large-scale cloud LLM that can handle multi-modal data sources. We develop a resource-constrained reinforcement learning (RCRL) strategy for TMO that optimizes the inference location (i.e., local vs. cloud) and multi-modal data sources to use for each task/dialogue, aiming to maximize the long-term reward (response quality, latency, and usage cost) while adhering to resource constraints. We also contribute M4A1, a new dataset we curated that contains reward and cost metrics across multiple modality, task, dialogue, and LLM configurations, enabling evaluation of offloading decisions. We demonstrate the effectiveness of TMO compared to several exploration-decision and LLM-as-Agent baselines, showing significant improvements in latency, cost, and response quality.

Yun-Wei Chu, Dong-Jun Han, Christopher G. Brinton

Federated learning (FL) is a promising distributed machine learning paradigm that enables multiple clients to collaboratively train a global model. In this paper, we focus on a practical federated multilingual learning setup where clients with their own language-specific data aim to collaboratively construct a high-quality neural machine translation (NMT) model. However, communication constraints in practical network systems present challenges for exchanging large-scale NMT engines between FL parties. We propose a meta-learning-based adaptive parameter selection methodology, MetaSend, that improves the communication efficiency of model transmissions from clients during FL-based multilingual NMT training. Our approach learns a dynamic threshold for filtering parameters prior to transmission without compromising the NMT model quality, based on the tensor deviations of clients between different FL rounds. Through experiments on two NMT datasets with different language distributions, we demonstrate that MetaSend obtains substantial improvements over baselines in translation quality in the presence of a limited communication budget.

Shahryar Zehtabi, Dong-Jun Han, Rohit Parasnis et al.

Decentralized federated learning (DFL) captures FL settings where both (i) model updates and (ii) model aggregations are exclusively carried out by the clients without a central server. Existing DFL works have mostly focused on settings where clients conduct a fixed number of local updates between local model exchanges, overlooking heterogeneity and dynamics in communication and computation capabilities. In this work, we propose Decentralized Sporadic Federated Learning ($\texttt{DSpodFL}$), a DFL methodology built on a generalized notion of $\textit{sporadicity}$ in both local gradient and aggregation processes. $\texttt{DSpodFL}$ subsumes many existing decentralized optimization methods under a unified algorithmic framework by modeling the per-iteration (i) occurrence of gradient descent at each client and (ii) exchange of models between client pairs as arbitrary indicator random variables, thus capturing $\textit{heterogeneous and time-varying}$ computation/communication scenarios. We analytically characterize the convergence behavior of $\texttt{DSpodFL}$ for both convex and non-convex models and for both constant and diminishing learning rates, under mild assumptions on the communication graph connectivity, data heterogeneity across clients, and gradient noises. We show how our bounds recover existing results from decentralized gradient descent as special cases. Experiments demonstrate that $\texttt{DSpodFL}$ consistently achieves improved training speeds compared with baselines under various system settings.

Liangqi Yuan, Dong-Jun Han, Christopher G. Brinton et al.

The rise of large language models (LLMs) has made natural language-driven route planning an emerging research area that encompasses rich user objectives. Current research exhibits two distinct approaches: direct route planning using LLM-as-Agent and graph-based searching strategies. However, LLMs in the former approach struggle to handle extensive map data, while the latter shows limited capability in understanding natural language preferences. Additionally, a more critical challenge arises from the highly heterogeneous and unpredictable spatio-temporal distribution of users across the globe. In this paper, we introduce a novel LLM-Assisted route Planning (LLMAP) system that employs an LLM-as-Parser to comprehend natural language, identify tasks, and extract user preferences and recognize task dependencies, coupled with a Multi-Step Graph construction with iterative Search (MSGS) algorithm as the underlying solver for optimal route finding. Our multi-objective optimization approach adaptively tunes objective weights to maximize points of interest (POI) quality and task completion rate while minimizing route distance, subject to three key constraints: user time limits, POI opening hours, and task dependencies. We conduct extensive experiments using 1,000 routing prompts sampled with varying complexity across 14 countries and 27 cities worldwide. The results demonstrate that our approach achieves superior performance with guarantees across multiple constraints.

Kyeongkook Seo, Dong-Jun Han, Jaejun Yoo

Despite recent advancements in federated learning (FL), the integration of generative models into FL has been limited due to challenges such as high communication costs and unstable training in heterogeneous data environments. To address these issues, we propose PRISM, a FL framework tailored for generative models that ensures (i) stable performance in heterogeneous data distributions and (ii) resource efficiency in terms of communication cost and final model size. The key of our method is to search for an optimal stochastic binary mask for a random network rather than updating the model weights, identifying a sparse subnetwork with high generative performance; i.e., a ``strong lottery ticket''. By communicating binary masks in a stochastic manner, PRISM minimizes communication overhead. This approach, combined with the utilization of maximum mean discrepancy (MMD) loss and a mask-aware dynamic moving average aggregation method (MADA) on the server side, facilitates stable and strong generative capabilities by mitigating local divergence in FL scenarios. Moreover, thanks to its sparsifying characteristic, PRISM yields a lightweight model without extra pruning or quantization, making it ideal for environments such as edge devices. Experiments on MNIST, FMNIST, CelebA, and CIFAR10 demonstrate that PRISM outperforms existing methods, while maintaining privacy with minimal communication costs. PRISM is the first to successfully generate images under challenging non-IID and privacy-preserving FL environments on complex datasets, where previous methods have struggled.

Wonjeong Choi, Jungwuk Park, Dong-Jun Han et al.

Research interests in the robustness of deep neural networks against domain shifts have been rapidly increasing in recent years. Most existing works, however, focus on improving the accuracy of the model, not the calibration performance which is another important requirement for trustworthy AI systems. Temperature scaling (TS), an accuracy-preserving post-hoc calibration method, has been proven to be effective in in-domain settings, but not in out-of-domain (OOD) due to the difficulty in obtaining a validation set for the unseen domain beforehand. In this paper, we propose consistency-guided temperature scaling (CTS), a new temperature scaling strategy that can significantly enhance the OOD calibration performance by providing mutual supervision among data samples in the source domains. Motivated by our observation that over-confidence stemming from inconsistent sample predictions is the main obstacle to OOD calibration, we propose to guide the scaling process by taking consistencies into account in terms of two different aspects -- style and content -- which are the key components that can well-represent data samples in multi-domain settings. Experimental results demonstrate that our proposed strategy outperforms existing works, achieving superior OOD calibration performance on various datasets. This can be accomplished by employing only the source domains without compromising accuracy, making our scheme directly applicable to various trustworthy AI systems.

Wenzhi Fang, Dong-Jun Han, Liangqi Yuan et al.

Device-cloud collaboration has emerged as a promising paradigm for deploying large language models (LLMs), combining the efficiency of lightweight on-device inference with the superior performance of powerful cloud LLMs. An essential problem in this scenario lies in deciding whether a given query is best handled locally or delegated to the cloud. Existing approaches typically rely on external routers, implemented as binary classifiers, which often struggle to determine task difficulty from the prompt's surface pattern. To address these limitations, we propose a framework where the on-device LLM makes routing decisions at the end of its solving process, with this capability instilled through post-training. In particular, we formulate a reward maximization problem with carefully designed rewards that encourage effective problem solving and judicious offloading to the cloud. To solve this problem, we develop a group-adaptive policy gradient algorithm, featuring a group-level policy gradient, designed to yield an unbiased gradient estimator of the reward, and adaptive prompt filtering, developed to enforce the constraint on cloud LLM usage. Extensive experiments across models and benchmarks show that the proposed methodology consistently outperforms existing baselines and significantly narrows the gap to full cloud LLM performance.

Guangchen Lan, Sipeng Zhang, Tianle Wang et al.

As the era of large language models (LLMs) on behalf of users unfolds, Preference Optimization (PO) methods have become a central approach to aligning LLMs with human preferences and improving performance. We propose Maximum a Posteriori Preference Optimization (MaPPO), a framework for learning from preferences that explicitly incorporates prior reward knowledge into the optimization objective. While existing methods such as Direct Preference Optimization (DPO) and its variants treat preference learning as a Maximum Likelihood Estimation (MLE) problem, MaPPO extends this paradigm by integrating prior reward estimates into a principled Maximum a Posteriori (MaP) objective. This not only generalizes DPO and its variants, but also enhances alignment by mitigating the oversimplified binary classification of responses. More importantly, MaPPO introduces no additional hyperparameter, and supports preference optimization in both offline and online settings. In addition, MaPPO can be used as a plugin with consistent improvement on DPO variants, including widely used SimPO, IPO, and CPO. Extensive empirical evaluations of different model sizes and model series on three standard benchmarks, including MT-Bench, AlpacaEval 2.0, and Arena-Hard, demonstrate consistent improvements in alignment performance without sacrificing computational efficiency.

Yun-Wei Chu, Dong-Jun Han, Seyyedali Hosseinalipour et al.

A few recent studies have demonstrated that leveraging centrally pre-trained models can offer advantageous initializations for federated learning (FL). However, existing pre-training methods do not generalize well when faced with an arbitrary set of downstream FL tasks. Specifically, they often (i) achieve limited average accuracy, particularly when there are unseen downstream labels, and (ii) result in significant accuracy variance, failing to provide a balanced performance across clients. To address these challenges, we propose CoPreFL, a collaborative/distributed pre-training approach which provides a robust initialization for downstream FL tasks. The key idea of CoPreFL is a model-agnostic meta-learning (MAML) procedure that tailors the global model to closely mimic heterogeneous and unseen FL scenarios, resulting in a pre-trained model that is rapidly adaptable to arbitrary FL tasks. Our MAML procedure incorporates performance variance into the meta-objective function, balancing performance across clients rather than solely optimizing for accuracy. Through extensive experiments, we demonstrate that CoPreFL obtains significant improvements in both average accuracy and variance across arbitrary downstream FL tasks with unseen/seen labels, compared with various pre-training baselines. We also show how CoPreFL is compatible with different well-known FL algorithms applied by the downstream tasks, enhancing performance in each case.

Liangqi Yuan, Chuhao Deng, Dong-Jun Han et al.

With the rapid advancement of Large Language Models (LLMs), their capabilities in various automation domains, particularly Unmanned Aerial Vehicle (UAV) operations, have garnered increasing attention. Current research remains predominantly constrained to small-scale UAV applications, with most studies focusing on isolated components such as path planning for toy drones, while lacking comprehensive investigation of medium- and long-range UAV systems in real-world operational contexts. Larger UAV platforms introduce distinct challenges, including stringent requirements for airport-based take-off and landing procedures, adherence to complex regulatory frameworks, and specialized operational capabilities with elevated mission expectations. This position paper presents the Next-Generation LLM for UAV (NeLV) system -- a comprehensive demonstration and automation roadmap for integrating LLMs into multi-scale UAV operations. The NeLV system processes natural language instructions to orchestrate short-, medium-, and long-range UAV missions through five key technical components: (i) LLM-as-Parser for instruction interpretation, (ii) Route Planner for Points of Interest (POI) determination, (iii) Path Planner for waypoint generation, (iv) Control Platform for executable trajectory implementation, and (v) UAV monitoring. We demonstrate the system's feasibility through three representative use cases spanning different operational scales: multi-UAV patrol, multi-POI delivery, and multi-hop relocation. Beyond the current implementation, we establish a five-level automation taxonomy that charts the evolution from current LLM-as-Parser capabilities (Level 1) to fully autonomous LLM-as-Autopilot systems (Level 5), identifying technical prerequisites and research challenges at each stage.

Shahryar Zehtabi, Dong-Jun Han, Seyyedali Hosseinalipour et al.

Much of federated learning (FL) focuses on settings where local dataset statistics remain the same between training and testing. However, this assumption often does not hold in practice due to distribution shifts, motivating the development of domain generalization (DG) approaches that leverage source domain data to train models capable of generalizing to unseen target domains. In this paper, we are motivated by two major gaps in existing work on FL and DG: (1) the lack of formal mathematical analysis of DG objectives; and (2) DG research in FL being limited to the star-topology architecture. We develop Decentralized Federated Domain Generalization with Style Sharing ($\textit{StyleDDG}$), a decentralized DG algorithm which allows devices in a peer-to-peer network to achieve DG based on sharing style information inferred from their datasets. Additionally, we provide the first systematic approach to analyzing style-based DG training in decentralized networks. We cast existing centralized DG algorithms within our framework, and employ their formalisms to model $\textit{StyleDDG}$. We then obtain analytical conditions under which convergence of $\textit{StyleDDG}$ can be guaranteed. Through experiments on popular DG datasets, we demonstrate that $\textit{StyleDDG}$ can obtain significant improvements in accuracy across target domains with minimal communication overhead compared to baseline decentralized gradient methods.

Evan Chen, Shiqiang Wang, Jianing Zhang et al.

In federated learning (FL), model training performance is strongly impacted by data heterogeneity across clients. Client-drift compensation methods have recently emerged as a solution to this issue, introducing correction terms into local model updates. To date, these methods have only been considered under stochastic gradient descent (SGD)-based model training, while modern FL frameworks also employ adaptive optimizers (e.g., Adam) for improved convergence. However, due to the complex interplay between first and second moments found in most adaptive optimization methods, naively injecting correction terms can lead to performance degradation in heterogeneous settings. In this work, we propose {\tt FAdamGC}, the first algorithm to integrate drift compensation into adaptive federated optimization. The key idea of {\tt FAdamGC} is injecting a pre-estimation correction term that aligns with the moment structure of adaptive methods. We provide a rigorous convergence analysis of our algorithm under non-convex settings, showing that {\tt FAdamGC} results in better rate and milder assumptions than naively porting SGD-based correction algorithms into adaptive optimizers. Our experimental results demonstrate that {\tt FAdamGC} consistently outperform existing methods in total communication and computation cost across varying levels of data heterogeneity, showing the efficacy of correcting gradient information in federated adaptive optimization.

Wenzhi Fang, Dong-Jun Han, Liangqi Yuan et al.

Fine-tuning large language models (LLMs) on resource-constrained clients remains a challenging problem. Recent works have fused low-rank adaptation (LoRA) techniques with federated fine-tuning to mitigate challenges associated with client model sizes and data scarcity. Still, the heterogeneity of resources remains a critical bottleneck: while higher-rank modules generally enhance performance, varying client capabilities constrain LoRA's feasible rank range. Existing approaches attempting to resolve this issue either lack analytical justification or impose additional computational overhead, leaving a wide gap for efficient and theoretically-grounded solutions. To address these challenges, we propose federated sketching LoRA (FSLoRA), which leverages a sketching mechanism to enable clients to selectively update submatrices of global LoRA modules maintained by the server. By adjusting the sketching ratios, which determine the ranks of the submatrices on the clients, FSLoRA flexibly adapts to client-specific communication and computational constraints. We provide a rigorous convergence analysis of FSLoRA that characterizes how the sketching ratios affect the convergence rate. Through comprehensive experiments on multiple datasets and LLM models, we demonstrate FSLoRA's performance improvements compared to various baselines.

Yongsheng Mei, Liangqi Yuan, Dong-Jun Han et al.

Federated learning (FL) has become a cornerstone in decentralized learning, where, in many scenarios, the incoming data distribution will change dynamically over time, introducing continuous learning (CL) problems. This continual federated learning (CFL) task presents unique challenges, particularly regarding catastrophic forgetting and non-IID input data. Existing solutions include using a replay buffer to store historical data or leveraging generative adversarial networks. Nevertheless, motivated by recent advancements in the diffusion model for generative tasks, this paper introduces DCFL, a novel framework tailored to address the challenges of CFL in dynamic distributed learning environments. Our approach harnesses the power of the conditional diffusion model to generate synthetic historical data at each local device during communication, effectively mitigating latent shifts in dynamic data distribution inputs. We provide the convergence bound for the proposed CFL framework and demonstrate its promising performance across multiple datasets, showcasing its effectiveness in tackling the complexities of CFL tasks.

Evan Chen, Frank Po-Chen Lin, Dong-Jun Han et al.

While federated learning (FL) eliminates the transmission of raw data over a network, it is still vulnerable to privacy breaches from the communicated model parameters. In this work, we propose Multi-Tier Federated Learning with Multi-Tier Differential Privacy (M^2FDP), a DP-enhanced FL methodology for jointly optimizing privacy and performance in hierarchical networks. One of the key concepts of M^2FDP is to extend the concept of HDP towards Multi-Tier Differential Privacy (MDP), while also adapting DP noise injection at different layers of an established FL hierarchy -- edge devices, edge servers, and cloud servers -- according to the trust models within particular subnetworks. We conduct a comprehensive analysis of the convergence behavior of M^2FDP, revealing conditions on parameter tuning under which the training process converges sublinearly to a finite stationarity gap that depends on the network hierarchy, trust model, and target privacy level. Subsequent numerical evaluations demonstrate that M^2FDP obtains substantial improvements in these metrics over baselines for different privacy budgets, and validate the impact of different system configurations.

Dong-Jun Han, Seyyedali Hosseinalipour, David J. Love et al.

While network coverage maps continue to expand, many devices located in remote areas remain unconnected to terrestrial communication infrastructures, preventing them from getting access to the associated data-driven services. In this paper, we propose a ground-to-satellite cooperative federated learning (FL) methodology to facilitate machine learning service management over remote regions. Our methodology orchestrates satellite constellations to provide the following key functions during FL: (i) processing data offloaded from ground devices, (ii) aggregating models within device clusters, and (iii) relaying models/data to other satellites via inter-satellite links (ISLs). Due to the limited coverage time of each satellite over a particular remote area, we facilitate satellite transmission of trained models and acquired data to neighboring satellites via ISL, so that the incoming satellite can continue conducting FL for the region. We theoretically analyze the convergence behavior of our algorithm, and develop a training latency minimizer which optimizes over satellite-specific network resources, including the amount of data to be offloaded from ground devices to satellites and satellites' computation speeds. Through experiments on three datasets, we show that our methodology can significantly speed up the convergence of FL compared with terrestrial-only and other satellite baseline approaches.

Beongjun Choi, Jy-yong Sohn, Dong-Jun Han et al.

Federated learning has been spotlighted as a way to train neural networks using distributed data with no need for individual nodes to share data. Unfortunately, it has also been shown that adversaries may be able to extract local data contents off model parameters transmitted during federated learning. A recent solution based on the secure aggregation primitive enabled privacy-preserving federated learning, but at the expense of significant extra communication/computational resources. In this paper, we propose a low-complexity scheme that provides data privacy using substantially reduced communication/computational resources relative to the existing secure solution. The key idea behind the suggested scheme is to design the topology of secret-sharing nodes as a sparse random graph instead of the complete graph corresponding to the existing solution. We first obtain the necessary and sufficient condition on the graph to guarantee both reliability and privacy. We then suggest using the Erdős-Rényi graph in particular and provide theoretical guarantees on the reliability/privacy of the proposed scheme. Through extensive real-world experiments, we demonstrate that our scheme, using only $20 \sim 30\%$ of the resources required in the conventional scheme, maintains virtually the same levels of reliability and data privacy in practical federated learning systems.

Jy-yong Sohn, Dong-Jun Han, Beongjun Choi et al.

Recent advances in large-scale distributed learning algorithms have enabled communication-efficient training via SignSGD. Unfortunately, a major issue continues to plague distributed learning: namely, Byzantine failures may incur serious degradation in learning accuracy. This paper proposes Election Coding, a coding-theoretic framework to guarantee Byzantine-robustness for SignSGD with Majority Vote, which uses minimum worker-master communication in both directions. The suggested framework explores new information-theoretic limits of finding the majority opinion when some workers could be malicious, and paves the road to implement robust and efficient distributed learning algorithms. Under this framework, we construct two types of explicit codes, random Bernoulli codes and deterministic algebraic codes, that can tolerate Byzantine attacks with a controlled amount of computational redundancy. For the Bernoulli codes, we provide upper bounds on the error probability in estimating the majority opinion, which give useful insights into code design for tolerating Byzantine attacks. As for deterministic codes, we construct an explicit code which perfectly tolerates Byzantines, and provide tight upper/lower bounds on the minimum required computational redundancy. Finally, the Byzantine-tolerance of the suggested coding schemes is confirmed by deep learning experiments on Amazon EC2 using Python with MPI4py package.