arXiv subjects

Arkajyoti Mitra

Publications and source records attributed to Arkajyoti Mitra.

3 recordsLinked to original sources

arXiv2025

FedVLM: Scalable Personalized Vision-Language Models through Federated Learning

Vision-language models (VLMs) demonstrate impressive zero-shot and few-shot learning capabilities, making them essential for several downstream tasks. However, fine-tuning these models at scale remains challenging, particularly in federated environments where data is decentralized and non-iid across clients. Existing parameter-efficient tuning methods like LoRA (Low-Rank Adaptation) reduce computational overhead but struggle with heterogeneous client data, leading to suboptimal generalization. To address these challenges, we propose FedVLM, a federated LoRA fine-tuning framework that enables decentralized adaptation of VLMs while preserving model privacy and reducing reliance on centralized training. To further tackle data heterogeneity, we introduce personalized LoRA (pLoRA), which dynamically adapts LoRA parameters to each client's unique data distribution, significantly improving local adaptation while maintaining global model aggregation. Experiments on the RLAIF-V dataset show that pLoRA improves client-specific performance by 24.5% over standard LoRA, demonstrating superior adaptation in non-iid settings. FedVLM provides a scalable and efficient solution for fine-tuning VLMs in federated settings, advancing personalized adaptation in distributed learning scenarios.

cs.CV↗

arXiv2025

Enhancing Graph Neural Networks: A Mutual Learning Approach

Knowledge distillation (KD) techniques have emerged as a powerful tool for transferring expertise from complex teacher models to lightweight student models, particularly beneficial for deploying high-performance models in resource-constrained devices. This approach has been successfully applied to graph neural networks (GNNs), harnessing their expressive capabilities to generate node embeddings that capture structural and feature-related information. In this study, we depart from the conventional KD approach by exploring the potential of collaborative learning among GNNs. In the absence of a pre-trained teacher model, we show that relatively simple and shallow GNN architectures can synergetically learn efficient models capable of performing better during inference, particularly in tackling multiple tasks. We propose a collaborative learning framework where ensembles of student GNNs mutually teach each other throughout the training process. We introduce an adaptive logit weighting unit to facilitate efficient knowledge exchange among models and an entropy enhancement technique to improve mutual learning. These components dynamically empower the models to adapt their learning strategies during training, optimizing their performance for downstream tasks. Extensive experiments conducted on three datasets each for node and graph classification demonstrate the effectiveness of our approach.

cs.LG↗

arXiv2025

Discovering New Shadow Patterns for Black-Box Attacks on Lane Detection of Autonomous Vehicles

We present a novel physical-world attack on autonomous vehicle (AV) lane detection systems that leverages negative shadows -- bright, lane-like patterns projected by passively redirecting sunlight through occluders. These patterns exploit intensity-based heuristics in modern lane detection (LD) algorithms, causing AVs to misclassify them as genuine lane markings. Unlike prior attacks, our method is entirely passive, power-free, and inconspicuous to human observers, enabling legal and stealthy deployment in public environments. Through simulation, physical testbed, and controlled field evaluations, we demonstrate that negative shadows can cause up to 100% off-road deviation or collision rates in specific scenarios; for example, a 20-meter shadow leads to complete off-road exits at speeds above 10 mph, while 30-meter shadows trigger consistent lane confusion and collisions. A user study confirms the attack's stealthiness, with 83.6% of participants failing to detect it during driving tasks. To mitigate this threat, we propose Luminosity Filter Pre-processing, a lightweight defense that reduces attack success by 87% through brightness normalization and selective filtering. Our findings expose a critical vulnerability in current LD systems and underscore the need for robust perception defenses against passive, real-world attacks.

cs.CR↗