A5068577161- Adversarial Robustness in Machine Learning
- Domain Adaptation and Few-Shot Learning
- Biometric Identification and Security
- Advanced Neural Network Applications
- AI in cancer detection
- User Authentication and Security Systems
- Advanced Steganography and Watermarking Techniques
- CRISPR and Genetic Engineering
- Digital Imaging for Blood Diseases
- RNA and protein synthesis mechanisms
- Radiomics and Machine Learning in Medical Imaging
- COVID-19 diagnosis using AI
- Hepatocellular Carcinoma Treatment and Prognosis
- Multimodal Machine Learning Applications
- Industrial Vision Systems and Defect Detection
- Nuclear Materials and Properties
- Video Coding and Compression Technologies
- Machine Learning and Data Classification
- Software System Performance and Reliability
- Generative Adversarial Networks and Image Synthesis
- Biomedical Text Mining and Ontologies
- Text and Document Classification Technologies
- Parasitic Diseases Research and Treatment
- Advanced biosensing and bioanalysis techniques
- Advanced Data Compression Techniques
Ghent University Global Campus
2021-2024
Ghent University
2023
Korea Advanced Institute of Science and Technology
2009-2011
Thanks to high-speed Internet access and feature-rich mobile devices, the demand for ubiquitous secure surveillance systems has increased. In this paper, we propose a privacy-protected video system that makes use of scalable coding (SVC). SVC can be used fulfill requirement omnipresence. Further, address privacy concerns, detect face regions subsequently scramble these regions-of-interest (ROIs) in compressed domain. To demonstrate feasibility proposed system, simulation results are...
Although supervised learning has been highly successful in improving the state-of-the-art domain of image-based computer vision past, margin improvement diminished significantly recent years, indicating that a plateau is sight. Meanwhile, use self-supervised (SSL) for purpose natural language processing (NLP) seen tremendous successes during past couple with this new paradigm yielding powerful models. Inspired by excellent results obtained field NLP, methods rely on clustering, contrastive...
Alarming increase in identity theft cases calls for the use of secure authentication systems that can clearly distinguish between authorized users and unauthorized who are possession valid security tokens or passwords. To this end, we propose a multi-factor system using user-specific pseudo-random numbers fingerprints to generate revocable privacy preserving biometric templates, which turn used authentication. We evaluated performance proposed on publicly available Fingerprint Verification...
Trypanosomiasis, a neglected tropical disease (NTD), challenges communities in sub-Saharan Africa and Latin America. The World Health Organization underscores the need for practical, field-adaptable diagnostics rapid screening tools to address negative impact of NTDs. While artificial intelligence has shown promising results screening, lack curated datasets impedes progress. In response this challenge, we developed Tryp dataset, comprising microscopy images unstained thick blood smears...
ImageNet, an influential dataset in computer vision, is traditionally evaluated using single-label classification, which assumes that image can be adequately described by a single concept or label. However, this approach may not fully capture the complex semantics within images available potentially hindering development of models effectively learn these intricacies. This study critically examines prevalent benchmarking and advocates for shift to multi-label ImageNet. would enable more...
CRISPR-Cas-Docker is a web server for in silico docking experiments with CRISPR RNAs (crRNAs) and Cas proteins. This aims at providing experimentalists the optimal crRNA-Cas pair predicted computationally when prokaryotic genomes have multiple arrays systems, as frequently observed metagenomic data.CRISPR-Cas-Docker provides two methods to predict protein given particular crRNA sequence: structure-based method (in docking) sequence-based (machine learning classification). For method, users...
Large-scale datasets for single-label multi-class classification, such as \emph{ImageNet-1k}, have been instrumental in advancing deep learning and computer vision. However, a critical often understudied aspect is the comprehensive quality assessment of these datasets, especially regarding potential multi-label annotation errors. In this paper, we introduce lightweight, user-friendly, scalable framework that synergizes human machine intelligence efficient dataset validation enhancement. We...
Machine learning (ML) research strongly relies on benchmarks in order to determine the relative effectiveness of newly proposed models. Recently, a number prominent effort argued that models improve state-of-the-art by small margin tend do so winning what they call "benchmark lottery". An important benchmark field machine and computer vision is ImageNet where are often showcased based their performance this dataset. Given large self-supervised (SSL) frameworks has been past couple years each...
Although the adoption rate of deep neural networks (DNNs) has tremendously increased in recent years, a solution for their vulnerability against adversarial examples not yet been found. As result, substantial research efforts are dedicated to fix this weakness, with many studies typically using subset source images generate examples, treating every image as equal. We demonstrate that, fact, is equally suited kind assessment. To do so, we devise large-scale model-to-model transferability...
Deep Neural Network (DNN) models are increasingly evaluated using new replication test datasets, which have been carefully created to be similar older and popular benchmark datasets. However, running counter expectations, DNN classification show significant, consistent, largely unexplained degradation in accuracy on these While the evaluation approach is assess of a model by making use all datapoints available respective we argue that doing so hinders us from adequately capturing behavior...
Abstract Motivation CRISPR-Cas-Docker is a web server for in silico docking experiments with CRISPR RNAs (crRNAs) and Cas proteins. This aims at providing experimentalists the optimal crRNA-Cas pair predicted computationally when prokaryotic genomes have multiple arrays systems, as frequently observed metagenomic data. provides two methods to predict protein given particular crRN sequence: structure-based method ( docking) sequence-based (machine learning classification). For method, users...