A5075464273- Advanced Adaptive Filtering Techniques
- Blind Source Separation Techniques
- Adversarial Robustness in Machine Learning
- Software Testing and Debugging Techniques
- Advanced Neural Network Applications
- Speech and Audio Processing
- Neural Networks and Applications
- Meat and Animal Product Quality
- Software Engineering Research
- Physiological and biochemical adaptations
- Anomaly Detection Techniques and Applications
- Chaos-based Image/Signal Encryption
- Cryptographic Implementations and Security
- Low-power high-performance VLSI design
- Ferroelectric and Negative Capacitance Devices
- Adipokines, Inflammation, and Metabolic Diseases
- Advanced Memory and Neural Computing
- Machine Learning in Bioinformatics
- Software Reliability and Analysis Research
- Probability and Statistical Research
- Advanced Malware Detection Techniques
- Neurobiology and Insect Physiology Research
- Numerical Methods and Algorithms
- Explainable Artificial Intelligence (XAI)
- IoT-based Smart Home Systems
Chosun University
2019-2024
The University of Texas at Austin
2019-2023
South China University of Technology
2020-2021
Iqra University
2016-2019
Usman Institute of Technology
2019
Mid Sweden University
2015
Cardiovascular Institute of the South
2012
University of Pennsylvania
2012
University of Pennsylvania Health System
2012
Translational Therapeutics (United States)
2012
The Internet of Things (IoT) being a promising technology the future is expected to connect billions devices. increased number communication generate mountains data and security can be threat. devices in architecture are essentially smaller size low powered. Conventional encryption algorithms generally computationally expensive due their complexity requires many rounds encrypt, wasting constrained energy gadgets. Less complex algorithm, however, may compromise desired integrity. In this...
Abstract Species living in extremely cold environments resist the freezing conditions through antifreeze proteins (AFPs). Apart from being essential for various organisms sub-zero temperatures, AFPs have numerous applications different industries. They possess very small resemblance to each other and cannot be easily identified using simple search algorithms such as BLAST PSI-BLAST. Diverse found fishes (Type I, II, III, IV glycoproteins (AFGPs)), are sub-types show low sequence structural...
The search for resonant mass bumps in invariant-mass distributions remains a cornerstone strategy uncovering Beyond the Standard Model (BSM) physics at Large Hadron Collider (LHC). Traditional methods often rely on predefined functional forms and exhaustive computational human resources, limiting scope of tested final states selections. This work presents BumpNet, machine learning-based approach leveraging advanced neural network architectures to generalize enhance Data-Directed Paradigm...
Gorham-Stout disease is an exceptionally rare which characterised by massive osteolysis of the bone, oedema, and in severe cases pleural effusion chylothorax. Its aetiopathology unknown, no specific treatment has been modulated thus far. We report case a 17-year-old male with bones his entire left arm persistent Due to late presentation patient’s desire for better quality life, amputation was only choice treatment. This evaluated treated at Orthopaedic Surgery Trauma department Rehman...
Automatic modulation classification (AMC) is a vital process in wireless communication systems that fundamentally problem. It employed to automatically determine the type of received signal. Deep learning (DL) methods have gained popularity addressing problem classification, as they learn features without needing technical expertise. However, their efficacy depends on complexity algorithm, which can be characterized by number parameters. In this research, we presented deep algorithm for AMC,...
In this paper, we propose an adaptive framework for the variable power of fractional least mean square (FLMS) algorithm. The proposed algorithm named as robust FLMS (RVP-FLMS) dynamically adapts to achieve high convergence rate with low steady state error. For evaluation purpose, problems system identification and channel equalization are considered. experiments clearly show that approach achieves better lower steady-state error compared FLMS. MATLAB code related simulation is available...
Antifreeze proteins (AFPs) are the sub-set of ice binding indispensable for species living in extreme cold weather. These bind to crystals, hindering their growth into large lattice that could cause physical damage. There variety AFPs found numerous organisms and due heterogeneous sequence characteristics, demonstrate a high degree diversity, which makes prediction challenging task. Herein, we propose machine learning framework deal with this vigorous diverse problem using manifolding...
In the recent years of industrial revolution, 3D printing has shown to grow as an expanding field new applications. The low cost solutions and short time market makes it a favorable candidate be utilized in dynamic fields engineering. Additive vast range applications many fields. This study presents wide printers along with comparison additive traditional manufacturing methods have been shown. A tutorial is presented explaining steps involved prototype using Rhinoceros Simplify software...
Due to the dynamic nature, chaotic time series are difficult predict. In conventional signal processing approaches signals treated either in or space domain only. Spatio-temporal analysis of provides more advantages over uni-dimensional by harnessing information from both temporal and spatial domains. Herein, we propose an spatio-temporal extension RBF neural networks for prediction series. The proposed algorithm utilizes concept time-space orthogonality separately deals with dynamics...
In this paper, we propose an adaptive framework for the variable power of fractional least mean square (FLMS) algorithm using concept instantaneous error energy. The proposed named power-FLMS (VP-FLMS) is computationally less expensive and dynamically adapts FLMS to achieve a high convergence rate with low steady state error. For evaluation purpose, problems channel estimation equalization are considered. experiments clearly show that approach achieves better lower steady-state compared FLMS.
In this paper, a fractional order calculus based least mean square algorithm is proposed for complex system identification. The algorithm, named as, (FCLMS), successfully deals with the problem of error due to negative weights or input/output in FLMS. For evaluation purpose linear considered. FCLMS identifies and achieve high convergence rate without compromising steady state error. two times better than that (CLMS).
Extracelluar matrix (ECM) proteins create complex networks of macromolecules which fill-in the extracellular spaces living tissues. They provide structural support and play an important role in maintaining cellular functions. Identification ECM can a vital studying various types diseases. Conventional wet lab–based methods are reliable; however, they expensive time consuming are, therefore, not scalable. In this research, we propose sequence-based novel machine learning approach for...
Deep Neural Network (DNN) inference demands substantial computing power, resulting in significant energy consumption. A large number of negative output activations convolution layers are rendered zero due to the invocation ReLU activation function. This results a unnecessary computations that consume amounts energy. paper presents ECHO: Energy-efficient Computation Harnessing Onilne Arithmetic - MSDF-based accelerator for DNN inference, designed computation pruning, utilizing an...
In this paper, we propose an adaptive framework for the variable step size of fractional least mean square (FLMS) algorithm. The proposed algorithm named robust size-FLMS (RVSS-FLMS), dynamically updates FLMS to achieve high convergence rate with low steady state error. For evaluation purpose, problem system identification is considered. experiments clearly show that approach achieves better compared and step-size modified (AMFLMS).
Deep neural network (DNN) inference demands substantial computing power, resulting in significant energy consumption. A large number of negative output activations convolution layers are rendered zero due to the invocation ReLU activation function. This results a unnecessary computations that consume amounts energy. paper presents ECHO, an accelerator for DNN designed computation pruning, utilizing unconventional arithmetic paradigm known as online/most digit first (MSDF) arithmetic, which...
Data conflict with regard to whether peroxisome proliferator-activated receptor-α agonism suppresses inflammation in humans. We hypothesized that healthy adults fenofibrate would blunt the induced immune responses endotoxin (lipopolysaccharide [LPS]), an vivo model for study of cardiometabolic inflammation.In Fenofibrate and omega-3 Fatty Acid Modulation Endotoxemia (FFAME) trial, 36 volunteers (mean age 26±7 years, mean body mass index 24±3 kg/m(2), 44% female, 72% white) were randomized...
Automatic modulation classification (AMC) is used to identify the for received signal. IoT devices use modern communication methods which are based on multiple input output (MIMO) in signals from various sources. The identification of vital. Feature AMC combined with deep learning techniques has potential meet latency requirement applications. An efficient convolutional neural network depthwise separable convolution been proposed classify signals. architecture 58% less parameters than...