arXiv subjects

Sina Najafi

Publications and source records attributed to Sina Najafi.

3 recordsLinked to original sources

arXiv2026

Joint Domain-Class Modeling for Federated Learning Under Feature Skew

Federated learning (FL) enables collaborative model training without centralizing private data, but performance often degrades under feature skew: clients share labels while the conditional input distributions $p_i(x\!\mid\!y)$ vary due to latent, client-specific appearance factors. We propose Joint Domain-Class Federated Learning (JDFL), a lightweight, optimizer-agnostic extension that makes this latent domain variation usable without sharing raw data. JDFL first infers domain clusters called pseudo-domains from brief local update signals. It then expands the classifier head to output $M\times C$, joint (domain-class) logits. This allows the model to represent domain-conditioned appearance while keeping a shared backbone. To train the expanded head we introduce two complementary supervision strategies based on simple intuitions: a similarity-aware soft-labeling that transfers evidence between nearby inferred domains while allowing domain-specific specialization, and a per-sample randomized target assignment that perturbs supervision across the joint outputs and serves as a low-cost training-time regularizer. JDFL integrates with existing standard FL methods (e.g., FedAvg, SCAFFOLD) with minimal changes. Empirically, both supervision modes consistently improve global test accuracy on standard domain-shifted image benchmarks; ablations and sensitivity studies show the gains stem from the proposed supervision and parametrization rather than mere capacity increase.

cs.LG↗

arXiv2026

FedRP: A Communication-Efficient Approach for Differentially Private Federated Learning Using Random Projection

Federated learning (FL) enables collaborative model training without centralizing data, but exchanging high-dimensional updates can expose sensitive information and incur substantial communication costs. We present FedRP, a communication-efficient method combining Gaussian random projection with consensus optimization based on the alternating direction method of multipliers (ADMM). In each round, clients project their model parameters into an $m$-dimensional space using a shared random matrix hidden from the server, which aggregates only compressed representations. We establish a high-probability guarantee linking projected-space consensus to proximity among client models and derive an $(ε,δ)$-differential privacy guarantee for each release under bounded $\ell_2$-sensitivity and a positive lower bound on parameter norms. With randomized parameter vectors and the projection matrix also limits information available to common reconstruction attacks. Experiments on MNIST and CIFAR-10 with LeNet-5 and a custom convolutional network show that FedRP achieves accuracy comparable to FedAvg and consistently exposes noise-perturbed privacy via FedAvg. Because clients transmit $m$ rather than $n$ values per round, FedRP reduces communication by orders of magnitude when $m \ll n$. The results demonstrate a favorable trade-off among accuracy, privacy, and communication efficiency. Code is available at https://github.com/mhnarimani/FedRP

cs.LG↗

arXiv2025

RadarSeq: A Temporal Vision Framework for User Churn Prediction via Radar Chart Sequences

Predicting user churn in non-subscription gig platforms, where disengagement is implicit, poses unique challenges due to the absence of explicit labels and the dynamic nature of user behavior. Existing methods often rely on aggregated snapshots or static visual representations, which obscure temporal cues critical for early detection. In this work, we propose a temporally-aware computer vision framework that models user behavioral patterns as a sequence of radar chart images, each encoding day-level behavioral features. By integrating a pretrained CNN encoder with a bidirectional LSTM, our architecture captures both spatial and temporal patterns underlying churn behavior. Extensive experiments on a large real-world dataset demonstrate that our method outperforms classical models and ViT-based radar chart baselines, yielding gains of 17.7 in F1 score, 29.4 in precision, and 16.1 in AUC, along with improved interpretability. The framework's modular design, explainability tools, and efficient deployment characteristics make it suitable for large-scale churn modeling in dynamic gig-economy platforms.

cs.CV↗