A5034055082- Privacy-Preserving Technologies in Data
- Cooperative Communication and Network Coding
- Advanced Image and Video Retrieval Techniques
- Sparse and Compressive Sensing Techniques
- Advanced Steganography and Watermarking Techniques
- Advanced MIMO Systems Optimization
- Advanced Wireless Communication Technologies
- Wireless Communication Security Techniques
- Radiomics and Machine Learning in Medical Imaging
- Medical Imaging Techniques and Applications
- Face and Expression Recognition
- Adversarial Robustness in Machine Learning
- Full-Duplex Wireless Communications
- Face recognition and analysis
- Neural Networks and Applications
- Biometric Identification and Security
- Peer-to-Peer Network Technologies
- Caching and Content Delivery
- Remote-Sensing Image Classification
- Internet Traffic Analysis and Secure E-voting
- Cryptography and Data Security
- Automated Road and Building Extraction
- Wireless Signal Modulation Classification
- Analog and Mixed-Signal Circuit Design
- Access Control and Trust
University of Geneva
2017-2024
Idiap Research Institute
2024
Harvard University
2021
Battelle
2019
Iran University of Science and Technology
2017
Ferdowsi University of Mashhad
2013-2015
Sadjad University of Technology
2012
Purpose The generalizability and trustworthiness of deep learning (DL)–based algorithms depend on the size heterogeneity training datasets. However, because patient privacy concerns ethical legal issues, sharing medical images between different centers is restricted. Our objective to build a federated DL-based framework for PET image segmentation utilizing multicentric dataset compare its performance with centralized DL approach. Methods from 405 head neck cancer patients 9 formed basis this...
Attenuation correction and scatter compensation (AC/SC) are two main steps toward quantitative PET imaging, which remain challenging in PET-only PET/MRI systems. These can be effectively tackled via deep learning (DL) methods. However, trustworthy, generalizable DL models commonly require well-curated, heterogeneous, large datasets from multiple clinical centers. At the same time, owing to legal/ethical issues privacy concerns, forming a collective, centralized dataset poses significant...
Generalizable and trustworthy deep learning models for PET/CT image segmentation require large heterogeneous multi-institutional datasets. However, legal, ethical, patient privacy issues challenge sharing of datasets between centers. To overcome these challenges, we developed a federated (FL) framework segmentation. A dataset consisting 328 FL (HN) cancer patients who underwent clinical examinations gathered from six different centers was enrolled. pure transformer network implemented as...
Image artefacts continue to pose challenges in clinical molecular imaging, resulting misdiagnoses, additional radiation doses patients and financial costs. Mismatch halo occur frequently gallium-68 (
In this paper, we propose a privacy preserving framework for outsourced media search applications. Considering three parties, data owner, clients and server, the owner out-sources description of his to an external which provides service on behalf owner. The proposed is based sparsifying transform with ambiguization, consists trained linear map, element-wise nonlinearity amplification. amplification technique makes it infeasible server learn structure database items queries. We demonstrate...
We assess the performance of a recurrent frame generation algorithm for prediction late frames from initial in dynamic brain PET imaging.Clinical 18 F-DOPA PET/CT studies 46 subjects with ten folds cross-validation were retrospectively employed. A novel stochastic adversarial video model was implemented to predict last 13 (25-90 minutes) (0-25 minutes). The quantitative analysis predicted performed test and validation dataset using established metrics.The images demonstrated that is capable...
Bottleneck problems are an important class of optimization that have recently gained increasing attention in the domain machine learning and information theory. They widely used generative models, fair algorithms, design privacy-assuring mechanisms, appear as information-theoretic performance bounds various multi-user communication problems. In this work, we propose a general family problems, termed <italic xmlns:mml="http://www.w3.org/1998/Math/MathML"...
In this paper, we consider a privacy preserving encoding framework for identification applications covering biometrics, physical object security and the Internet of Things (IoT). The proposed is based on sparsifying transform, which consists trained linear map, an element-wise nonlinearity, amplification. transform amplification are not symmetric data owner user. We demonstrate that approach closely related to sparse ternary codes (STC), recent information-theoretic concept fast approximate...
Abstract Background Notwithstanding the encouraging results of previous studies reporting on efficiency deep learning (DL) in COVID‐19 prognostication, clinical adoption developed methodology still needs to be improved. To overcome this limitation, we set out predict prognosis a large multi‐institutional cohort patients with using DL‐based model. Purpose This study aimed evaluate performance privacy‐preserving federated (DPFL) predicting outcomes chest CT images. Methods After applying...
In this study, we harness the information-theoretic Privacy Funnel (PF) model to develop a method for privacy-preserving representation learning using an end-to-end training framework. We rigorously address trade-off between obfuscation and utility. Both are quantified through logarithmic loss, measure also recognized as self-information loss. This exploration deepens interplay privacy learning, offering substantive insights into data protection mechanisms both discriminative generative...
In this paper we propose a novel relay selection method for cooperation communication networks using fuzzy logic. Many efforts have been made in the literature to select superior based on relay's SNR/SER and / or reputation (in stimulation methods). We jointly consider four criteria process of selection, SNR/SER, reputation, relaying strategy location. condition which network users employ different strategies, i.e., some nodes decode-and-forward strategy, amplify-and-forward other ones...
Some unique attributes of P2P networks such as cost efficiency and scalability, contributed for the widespread adaptation these networks. Since applications are mostly used in file-sharing, preserving anonymity users has become a very important subject researchers. As result, lot methods suggested to preserve users. Most methods, by relying on established anonymous solutions client/server applications, presented unstructured But structured overlays, using Distributed Hash Tables (DHT) their...
The energy of all sensor nodes in wireless networks is limited. For this reason, providing a method communication between and network administrator to manage consumption crucial. In paper, centralized evolutionary clustering protocol for proposed. proposed selects the appropriate as CHs according three criteria that ultimately increases life time. This paper investigates genetic algorithm (GA) dynamic technique find optimum CHs. Furthermore, an innovative fitness function Each chromosome...
In this paper, we propose a new relay selection scheme for multi-user cooperation communications using novel multiple criteria decision making optimization method. We consider jointly three in our proposed method, two of which have been studied and employed the process literature, separately, next one has not considered schemes. paper these as scheme. To best knowledge, no prior work from point view. Also, approach selecting superior candidate relay, i.e., an information theoretic measure...
We consider the problem of privacy-preserving data release for a specific utility task under perfect obfuscation constraint. establish necessary and sufficient condition to extract features original that carry as much information about attribute possible, while not revealing any sensitive attribute. This formulation generalizes both bottleneck privacy funnel problems. adopt local geometry analysis provides useful insight into coupling trajectory construction spherical perturbation...
Peer-to-Peer (P2P) networks are powerful distributed solutions which do not rely on external servers and gather required resources from their users. Therefore, fairness is an important feature for designers of these networks. Prevention free-riding a key aspect fairness. One the most file-sharing protocols in P2P BitTorrent. This protocol distinguished other systems its unique way uploading process novel approach to For first time by implementing eminent game theory strategy, "tit-for-tat"...
The performance of a relay-based cooperative cellular network is affected by the relay location. Many efforts have been made in literature for optimal placement from various perspectives and with different assumptions. A small group research work focuses on maximizing coverage region relaying networks finding location within network. In this paper, we propose simple selection technique objective We consider condition which two types relays network, fixed unfixed relays. are stations...
We study and analyze coverage region in MIMO communication systems for a multiple-relay network with decode-and-forward (DF) strategy at the relays. Assuming that there is line-of-sight (LOS) propagation environment source-relay channels channel state information available receivers (CSIR), we consider objective of maximizing given transmission rate show numerically significant effect on capacity bounds, optimal relay location region. Also, situation which two adjacent relays cooperate...
In this paper we study and analyze coverage region for half-duplex multiple-input multiple-output (MIMO) relay channel with amplify-and-forward (AF) strategy at the station. By assuming mixed Rayleigh Rician fading channels two different station situations, consider objective of maximizing a given transmission rate find optimal location in sense region. Using Monte Carlo simulations, capacity bounds are shown cases locations. Finally, compare our results previous ones obtained...
We propose a new computationally efficient privacy-preserving identification framework based on layered sparse coding. The key idea of the proposed is sparsifying transform learning with ambiguization, which consists trained linear map, component-wise nonlinearity and privacy amplification. introduce practical framework, two phases: public private identification. untrusted server provides fast search service protected codebook stored at its side. trusted or local client application performs...
In this paper, first we propose a new approach for mathematical multiple criteria decision making (MCDM) methods using information theoretic measures, entropy and divergence. Using the concept of entropy, determine impact each criterion in process. The Shannon's has been previously employed purpose. paper use Renyi's potential weight assessment. Next, introduce divergence as separation measure MCDM methods. results indicate that outperforms conventional Euclidean distance measure. These...
This paper investigates the usage of Conventional PID controller, Type-1 and Type-2 fuzzy controller in controlling liquid level three tank control system. All real systems exhibit non-linear nature, thus conventional controllers are not always able to provide good accurate results. Fuzzy logic (FLC) can be used obtain more precise In this a model for simulation is designed all assumptions made before development model. Analysis done through computer using Matlab/Simulink toolbox. Moreover,...