arXiv subjects

Jaemin Shin

Publications and source records attributed to Jaemin Shin.

12 recordsLinked to original sources

arXiv2026

Multimodal Federated Learning under Dual-Axis Modality Missingness

Multimodal federated learning (FL) supports collaborative modeling in privacy-sensitive health-sensing and medical settings, but realistic deployments often exhibit dual-axis modality missingness: clients have different modality sets, and individual samples may contain only subsets of the modalities available locally. Existing methods typically address these two axes separately. We propose Flux, a multimodal federated learning framework built around two complementary components. First, modality-aware confidence tempering learns sample-specific confidence for each modality through mask-aware unimodal supervision and fuses the confidence estimates from observed modalities into a sample-adaptive temperature that adjusts predictive sharpness according to evidence quality and completeness. Second, gradient-decoupled private adaptation applies this temperature only to a client-private prediction pathway, while training the shared federated model with a standard, untempered objective. This enables sample-specific, client-local confidence adaptation without allowing confidence-dependent gradients to perturb shared representation learning. Across four multimodal datasets, Flux achieves the highest average macro-F1 on every dataset, outperforming the strongest dataset-specific baseline by 0.8~2.2 points and by 1.6 points on average. Additional analyses demonstrate favorable calibration, temperature sensitivity to both modality missingness and input corruption, and more stable shared optimization under private-only tempering. Our code is available at https://github.com/AdibaOrz/Flux.

cs.LG↗

arXiv2024

(FL)$^2$: Overcoming Few Labels in Federated Semi-Supervised Learning

Federated Learning (FL) is a distributed machine learning framework that trains accurate global models while preserving clients' privacy-sensitive data. However, most FL approaches assume that clients possess labeled data, which is often not the case in practice. Federated Semi-Supervised Learning (FSSL) addresses this label deficiency problem, targeting situations where only the server has a small amount of labeled data while clients do not. However, a significant performance gap exists between Centralized Semi-Supervised Learning (SSL) and FSSL. This gap arises from confirmation bias, which is more pronounced in FSSL due to multiple local training epochs and the separation of labeled and unlabeled data. We propose $(FL)^2$, a robust training method for unlabeled clients using sharpness-aware consistency regularization. We show that regularizing the original pseudo-labeling loss is suboptimal, and hence we carefully select unlabeled samples for regularization. We further introduce client-specific adaptive thresholding and learning status-aware aggregation to adjust the training process based on the learning progress of each client. Our experiments on three benchmark datasets demonstrate that our approach significantly improves performance and bridges the gap with SSL, particularly in scenarios with scarce labeled data.

cs.LG↗

arXiv2024

Federated Learning for Time-Series Healthcare Sensing with Incomplete Modalities

Many healthcare sensing applications utilize multimodal time-series data from sensors embedded in mobile and wearable devices. Federated Learning (FL), with its privacy-preserving advantages, is particularly well-suited for health applications. However, most multimodal FL methods assume the availability of complete modality data for local training, which is often unrealistic. Moreover, recent approaches tackling incomplete modalities scale poorly and become inefficient as the number of modalities increases. To address these limitations, we propose FLISM, an efficient FL training algorithm with incomplete sensing modalities while maintaining high accuracy. FLISM employs three key techniques: (1) modality-invariant representation learning to extract effective features from clients with a diverse set of modalities, (2) modality quality-aware aggregation to prioritize contributions from clients with higher-quality modality data, and (3) global-aligned knowledge distillation to reduce local update shifts caused by modality differences. Extensive experiments on real-world datasets show that FLISM not only achieves high accuracy but is also faster and more efficient compared with state-of-the-art methods handling incomplete modality problems in FL. We release the code as open-source at https://github.com/AdibaOrz/FLISM.

cs.LG↗

arXiv2024

Not All Federated Learning Algorithms Are Created Equal: A Performance Evaluation Study

Federated Learning (FL) emerged as a practical approach to training a model from decentralized data. The proliferation of FL led to the development of numerous FL algorithms and mechanisms. Many prior efforts have given their primary focus on accuracy of those approaches, but there exists little understanding of other aspects such as computational overheads, performance and training stability, etc. To bridge this gap, we conduct extensive performance evaluation on several canonical FL algorithms (FedAvg, FedProx, FedYogi, FedAdam, SCAFFOLD, and FedDyn) by leveraging an open-source federated learning framework called Flame. Our comprehensive measurement study reveals that no single algorithm works best across different performance metrics. A few key observations are: (1) While some state-of-the-art algorithms achieve higher accuracy than others, they incur either higher computation overheads (FedDyn) or communication overheads (SCAFFOLD). (2) Recent algorithms present smaller standard deviation in accuracy across clients than FedAvg, indicating that the advanced algorithms' performances are stable. (3) However, algorithms such as FedDyn and SCAFFOLD are more prone to catastrophic failures without the support of additional techniques such as gradient clipping. We hope that our empirical study can help the community to build best practices in evaluating FL algorithms.

cs.LG↗

arXiv2024

Flame: Simplifying Topology Extension in Federated Learning

Distributed machine learning approaches, including a broad class of federated learning (FL) techniques, present a number of benefits when deploying machine learning applications over widely distributed infrastructures. The benefits are highly dependent on the details of the underlying machine learning topology, which specifies the functionality executed by the participating nodes, their dependencies and interconnections. Current systems lack the flexibility and extensibility necessary to customize the topology of a machine learning deployment. We present Flame, a new system that provides flexibility of the topology configuration of distributed FL applications around the specifics of a particular deployment context, and is easily extensible to support new FL architectures. Flame achieves this via a new high-level abstraction Topology Abstraction Graphs (TAGs). TAGs decouple the ML application logic from the underlying deployment details, making it possible to specialize the application deployment with reduced development effort. Flame is released as an open source project, and its flexibility and extensibility support a variety of topologies and mechanisms, and can facilitate the development of new FL methodologies.

cs.LG↗

arXiv2023

FedTherapist: Mental Health Monitoring with User-Generated Linguistic Expressions on Smartphones via Federated Learning

Psychiatrists diagnose mental disorders via the linguistic use of patients. Still, due to data privacy, existing passive mental health monitoring systems use alternative features such as activity, app usage, and location via mobile devices. We propose FedTherapist, a mobile mental health monitoring system that utilizes continuous speech and keyboard input in a privacy-preserving way via federated learning. We explore multiple model designs by comparing their performance and overhead for FedTherapist to overcome the complex nature of on-device language model training on smartphones. We further propose a Context-Aware Language Learning (CALL) methodology to effectively utilize smartphones' large and noisy text for mental health signal sensing. Our IRB-approved evaluation of the prediction of self-reported depression, stress, anxiety, and mood from 46 participants shows higher accuracy of FedTherapist compared with the performance with non-language features, achieving 0.15 AUROC improvement and 8.21% MAE reduction.

cs.CL↗

arXiv2022

FedBalancer: Data and Pace Control for Efficient Federated Learning on Heterogeneous Clients

Federated Learning (FL) trains a machine learning model on distributed clients without exposing individual data. Unlike centralized training that is usually based on carefully-organized data, FL deals with on-device data that are often unfiltered and imbalanced. As a result, conventional FL training protocol that treats all data equally leads to a waste of local computational resources and slows down the global learning process. To this end, we propose FedBalancer, a systematic FL framework that actively selects clients' training samples. Our sample selection strategy prioritizes more "informative" data while respecting privacy and computational capabilities of clients. To better utilize the sample selection to speed up global training, we further introduce an adaptive deadline control scheme that predicts the optimal deadline for each round with varying client training data. Compared with existing FL algorithms with deadline configuration methods, our evaluation on five datasets from three different domains shows that FedBalancer improves the time-to-accuracy performance by 1.20~4.48x while improving the model accuracy by 1.1~5.0%. We also show that FedBalancer is readily applicable to other FL approaches by demonstrating that FedBalancer improves the convergence speed and accuracy when operating jointly with three different FL algorithms.

cs.LG↗

arXiv2021

Necessary conditions for blow-up solutions to the restricted Euler--Poisson equations

In this work, we study the behavior of blow-up solutions to the multidimensional restricted Euler--Poisson equations which are the localized version of the full Euler--Poisson system. We provide necessary conditions for the existence of finite-time blow-up solutions in terms of the initial data, and describe the asymptotic behavior of the solutions near blow up times. We also identify a rich set of the initial data which yields global bounded solutions.

math.AP↗

arXiv2015

Spectral properties of the Neumann-Poincaré operator and uniformity of estimates for the conductivity equation with complex coefficients

We consider well-posedness of the boundary value problem in presence of an inclusion with complex conductivity $k$. We first consider the transmission problem in $\mathbb{R}^d$ and characterize solvability of the problem in terms of the spectrum of the Neumann-Poincaré operator. We then deal with the boundary value problem and show that the solution is bounded in its $H^1$-norm uniformly in $k$ as long as $k$ is at some distance from a closed interval in the negative real axis. We then show with an estimate that the solution depends on $k$ in its $H^1$-norm Lipschitz continuously. We finally show that the boundary perturbation formula in presence of a diametrically small inclusion is valid uniformly in $k$ away from the closed interval mentioned before. The results for the single inclusion case are extended to the case when there are multiple inclusions with different complex conductivities: We first obtain a complete characterization of solvability when inclusions consist of two disjoint disks and then prove solvability and uniform estimates when imaginary parts of conductivities have the same signs. The results are obtained using the spectral property of the associated Neumann-Poincaré operator and the spectral resolution.

math.AP↗

arXiv2015

Higher order operator splitting Fourier spectral methods for the Allen-Cahn equation

The Allen-Cahn equation is solved numerically by operator splitting Fourier spectral methods. The basic idea of the operator splitting method is to decompose the original problem into sub-equations and compose the approximate solution of the original equation using the solutions of the subproblems. Unlike the first and the second order methods, each of the heat and the free-energy evolution operators has at least one backward evaluation in higher order methods. We investigate the effect of negative time steps on a general form of third order schemes and suggest three third order methods for better stability and accuracy. Two fourth order methods are also presented. The traveling wave solution and a spinodal decomposition problem are used to demonstrate numerical properties and the order of convergence of the proposed methods.

math.NA↗

arXiv2010

The Cauchy-Dirichlet problem for the FENE dumbbell model of polymeric fluids

The FENE dumbbell model consists of the incompressible Navier-Stokes equation and the Fokker-Planck equation for the polymer distribution. In such a model, the polymer elongation cannot exceed a limit $\sqrt{b}$, yielding all interesting features near the boundary. In this paper we establish the local well-posedness for the FENE dumbbell model under a class of Dirichlet-type boundary conditions dictated by the parameter $b$. As a result, for each $b>0$ we identify a sharp boundary requirement for the underlying density distribution, while the sharpness follows from the existence result for each specification of the boundary behavior. It is shown that the probability density governed by the Fokker-Planck equation approaches zero near boundary, necessarily faster than the distance function $d$ for $b>2$, faster than $d|ln d|$ for $b=2$, and as fast as $d^{b/2}$ for $0<b<2$. Moreover, the sharp boundary requirement for $b\geq 2$ is also sufficient for the distribution to remain a probability density.

math.AP↗

arXiv2010

Global Well-Posedness for the Microscopic FENE Model with a Sharp Boundary Condition

We prove global well-posedness for the microscopic FENE model under a sharp boundary requirement. The well-posedness of the FENE model that consists of the incompressible Navier-Stokes equation and the Fokker-Planck equation has been studied intensively, mostly with the zero flux boundary condition. Recently it was illustrated by C. Liu and H. Liu [2008, SIAM J. Appl. Math., 68(5):1304--1315] that any preassigned boundary value of a weighted distribution will become redundant once the non-dimensional parameter $b>2$. In this article, we show that for the well-posedness of the microscopic FENE model ($b>2$) the least boundary requirement is that the distribution near boundary needs to approach zero faster than the distance function. Under this condition, it is shown that there exists a unique weak solution in a weighted Sobolev space. Moreover, such a condition still ensures that the distribution is a probability density. The sharpness of this boundary requirement is shown by a construction of infinitely many solutions when the distribution approaches zero as fast as the distance function.

math.AP↗