arXiv subjects

Javad Parsa

Publications and source records attributed to Javad Parsa.

4 recordsLinked to original sources

arXiv2026

SeqLoRA: Bilevel Orthogonal Adaptation for Continual Multi-Concept Generation

Parameter-efficient fine-tuning enables fast personalization of text-to-image diffusion models, but composing multiple custom concepts remains challenging due to representation interference. Existing modular methods either rely on expensive post-hoc fusion or freeze adaptation subspaces, which limit expressiveness and concept fidelity. To address this trade-off, we propose Sequential regularized LoRA (SeqLoRA), a constrained continual learning framework that jointly optimizes both LoRA factors via bilevel optimization. Theoretically, we establish strong convergence guarantees for our algorithm and model the residual layer activations as a matrix sub-Gaussian process to derive high-probability bounds on catastrophic forgetting. We further prove that learning the LoRA basis from data minimizes residual interference energy more effectively than frozen-basis methods. Experiments on multi-concept image generation demonstrate that SeqLoRA improves identity preservation and scalability across up to 101 concepts, while avoiding costly fusion and reducing attribute interference in composed generations.

cs.LG↗

arXiv2025

Byzantine-Robust Federated Learning with Learnable Aggregation Weights

Federated Learning (FL) enables clients to collaboratively train a global model without sharing their private data. However, the presence of malicious (Byzantine) clients poses significant challenges to the robustness of FL, particularly when data distributions across clients are heterogeneous. In this paper, we propose a novel Byzantine-robust FL optimization problem that incorporates adaptive weighting into the aggregation process. Unlike conventional approaches, our formulation treats aggregation weights as learnable parameters, jointly optimizing them alongside the global model parameters. To solve this optimization problem, we develop an alternating minimization algorithm with strong convergence guarantees under adversarial attack. We analyze the Byzantine resilience of the proposed objective. We evaluate the performance of our algorithm against state-of-the-art Byzantine-robust FL approaches across various datasets and attack scenarios. Experimental results demonstrate that our method consistently outperforms existing approaches, particularly in settings with highly heterogeneous data and a large proportion of malicious clients.

cs.LG↗

arXiv2025

Subject grounding to reduce electromagnetic interference for MRI scanners operating in unshielded environments

Purpose. Portable low-field (< 0.1 T) MRI is increasingly used for point-of-care imaging, but electromagnetic interference (EMI) presents a significant challenge, especially in unshielded environments. EMI can degrade image quality and compromise diagnostic utility. This study investigates whether subject grounding can effectively reduce EMI and improve image quality, comparing different grounding strategies. Methods. Experiments were conducted using a 47 mT Halbach-based MRI scanner with a single receive channel. Reproducibility was evaluated at a second site using a 72 mT scanner with similar geometry. Turbo spin echo sequences were used to image the hand and brain. Subject grounding was implemented using conductive cloth sleeves or ECG electrodes, each connected between the subject's skin and scanner ground. Three EMI conditions were tested: ambient, added single-frequency EMI, and broadband EMI. Signal-to-noise ratios (SNR) were calculated under each configuration. Results. Subject grounding significantly reduced EMI in both hand and brain scans. In hand imaging, conductive sleeves reduced noise from 85x to 1.25x the 50-ohm noise floor. In brain imaging, grounding alone reduced noise from 55x to 25x baseline; when combined with arc RF shields, noise was further reduced to 1.2x baseline, even under added EMI. These results were reproducible across different scanners and locations. Conclusion. Subject grounding is a simple, effective, and reproducible strategy for mitigating EMI in portable low-field MRI. It is especially effective for hand imaging, while brain imaging benefits from additional RF shielding. The approach is robust under various EMI conditions and may complement other denoising techniques.

physics.med-ph↗

arXiv2024

Balancing Application Relevant and Sparsity Revealing Excitation in Input Design

The maximum absolute correlation between regressors, which is called mutual coherence, plays an essential role in sparse estimation. A regressor matrix whose columns are highly correlated may result from optimal input design, since there is no constraint on the mutual coherence, making it difficult to handle sparse estimation. This paper aims to tackle this issue for fixed denominator models, which include Laguerre, Kautz, and generalized orthonormal basis function expansion models, for example. The paper proposes an optimal input design method where the achieved Fisher information matrix is fitted to the desired Fisher matrix, together with a coordinate transformation designed to make the regressors in the transformed coordinates have low mutual coherence. The method can be used together with any sparse estimation method and any desired Fisher matrix. A numerical study shows its potential for alleviating the problem of model order selection when used in conjunction with, for example, classical methods such as the Akaike Information Criterion.

eess.SY↗