J. S. Kumar

XTrust Vision An Explanation-Guided Reliability Assessment Framework for Trustworthy Vision Transformers

ViTs (Vision Transformers) have been developed as many of the most popular architectures in contemporary computer vision. They harness self-attention strategies to capture long-range dependencies of images. This allows them to achieve learning of rich global contextual information that is very challenging to model with...

Efficient Posterior Approximation for Bayesian Neural Networks in Large-Scale Distributed Environments

Bayesian neural networks (BNNs) introduce a principled framework for uncertainty estimation through learning probability distributions with respect to model parameters. Still, current deep networks are not only unfeasible to perform exact Bayesian inference, but traditional posterior approximation methods also incur si...

Auditable Explainable Threat Intelligence for IoT Networks using Interpretable Machine Learning and Blockchain

XTrust-LM: An Explanation-Guided Reliability Assessment Framework for Multilingual Retrieval-Augmented Language Models

Retrieval-Augmented Generation (RAG) can help augment Large Language Models (LLMs) with knowledge of the external context when generating and beyond response. But old-line RAG systems are limited in their response reliability, evidence consistency, explainability, and multilingual performance with significant user trus...

Autonomous Digital Twin Framework for Predictive Cyber Defense in Edge-IoT Networks

Cross-Model Attention-based Vision–Language Fusion for Interpretable Multi-Label Chest Disease Diagnosis

Chest diseases including pneumonia, cardiomegaly, pleural effusion, atelectasis and consolidation are major public health problems around the world that need to be diagnosed quickly and accurately to provide appropriate treatment. In this paper, we propose a cross-modal attention-based multimodal deep learning method t...

Quantum Support Vector Machine for Encrypted Botnet Command and Control Traffic Detection Without Payload Inspection

Modern botnets use encrypted Command-and-Control (C2) communications, which poses a major challenge for network security given the fact that payloads are encrypted, which makes the use of traditional signature-based inspection methods ineffective. In this work, we introduce QSVM Shield, a hybrid quantum-classical solut...