Article
Computer Science, Hardware & Architecture
Amin Nezarat, N. Seifadini
Summary: The study focuses on predicting trip mode distribution based on raw GPS trajectory data using a convolutional neural network (CNN) architecture. The key innovation lies in designing the input layer layout and normalization operation to be compatible with CNN architecture and represent essential motion features. The proposed configuration achieved a highest prediction accuracy of 85.26%.
JOURNAL OF SUPERCOMPUTING
(2021)
Article
Mechanics
Indu Kant Deo, Rajeev Jaiman
Summary: In this paper, a deep learning technique for data-driven predictions of wave propagation is presented. The technique utilizes an attention-based convolutional recurrent autoencoder network (AB-CRAN) to construct a low-dimensional representation of wave propagation data. The proposed AB-GRAN framework accurately captures and preserves the wave characteristics of the solution for long time horizons, outperforming the standard recurrent neural network.
Article
Biochemical Research Methods
Fei Zhu, Lei Deng, Yuhao Dai, Guangyu Zhang, Fanwang Meng, Cheng Luo, Guang Hu, Zhongjie Liang
Summary: In this study, a novel deep neural network PPICT was proposed to predict PTM cross-talk by combining protein sequence-structure-dynamics information and structural information for PPI graph. The study found that cross-talk events preferentially occur among residues with high co-evolution and high potential in allosteric regulation. The PPICT method significantly improves the prediction performance and can identify PTM cross-talk between proteins at the proteome level, providing insights into cross-talk between different signal pathways introduced by PTMs.
BRIEFINGS IN BIOINFORMATICS
(2023)
Article
Automation & Control Systems
Shengli Zhang, Xinjie Li
Summary: Therapeutic peptides play a crucial role in the functioning of various cell activities in an organism, and predicting these peptides is essential for peptide-based therapy research. We have developed a deep learning model called Pep-CNN, which accurately predicts therapeutic peptides.
CHEMOMETRICS AND INTELLIGENT LABORATORY SYSTEMS
(2022)
Article
Computer Science, Artificial Intelligence
Arun Singh Bhadwal, Kamal Kumar, Neeraj Kumar
Summary: This paper proposes a new representation called GenSMILES to overcome the limitations of SMILES representation and improve the validity and diversity of generated molecules. GenSMILES relies on derivative rules to address syntactical and semantic issues, and it allows for more efficient generation of valid molecules.
KNOWLEDGE-BASED SYSTEMS
(2023)
Article
Mechanics
Bo -Wen Xu, Sang Ye, Min Li, Hong -Ping Zhao, Xi-Qiao Feng
Summary: In this paper, a micromechanics-based deep learning method is proposed for predicting the strength of two-dimensional microcracked brittle materials. By generating a large dataset of microcracked specimens and their load-bearing capacity, a deep neural network is constructed. The experimental results show that the trained network can accurately and efficiently predict the load-bearing capacity of microcracked brittle materials.
ENGINEERING FRACTURE MECHANICS
(2022)
Article
Chemistry, Multidisciplinary
Yan Gao, Xiangjiu Che, Huan Xu, Mei Bie
Summary: The major challenges for medical image segmentation tasks are complex backgrounds and fuzzy boundaries. In order to reduce their negative impacts, an enhanced feature extraction network (EFEN) based on U-Net is proposed. The network utilizes feature re-extraction and improved skip-connection with positional encoding and a cross-attention mechanism to capture positional and relative information between organs while strengthening useful information and weakening noise. Experimental results show that EFEN outperforms U-Net and recent networks, achieving significant improvements in DSC on various datasets.
APPLIED SCIENCES-BASEL
(2023)
Article
Multidisciplinary Sciences
Xiaoying Su, Yanfeng Sun, Hongxi Liu, Qiuling Lang, Yichen Zhang, Jiquan Zhang, Chaoyong Wang, Yanan Chen
Summary: Nowadays, accurate prediction of diseases is crucial to alleviate the burden on the healthcare system. This paper proposes an innovative deep learning model, WOCLSA, which combines ANN, CNN, and LSTM models to improve disease prediction performance. Through extensive studies and tests, the results show that our prediction models achieve an accuracy of 91% in diagnosing COVID-19 infection, surpassing comparable models.
SCIENTIFIC REPORTS
(2023)
Article
Energy & Fuels
Jichong Lei, Chao Yang, Changan Ren, Wei Li, Chengwei Liu, Aikou Sun, Yukun Li, Zhenping Chen, Tao Yu
Summary: The DRAGON code is used to generate samples for developing a deep neural network-based nuclide density prediction model. The model shows lower prediction errors in both the low-burnup and high-burnup regions, demonstrating the feasibility of artificial intelligence in nuclide density prediction.
INTERNATIONAL JOURNAL OF ENERGY RESEARCH
(2022)
Article
Metallurgy & Metallurgical Engineering
Liu Hui, Deng Da-hua
Summary: This research proposes an enhanced hybrid ensemble deep learning model for PM2.5 forecasting and demonstrates its accuracy and effectiveness through experiments.
JOURNAL OF CENTRAL SOUTH UNIVERSITY
(2022)
Article
Water Resources
Gege Wen, Zongyi Li, Kamyar Azizzadenesheli, Anima Anandkumar, Sally M. Benson
Summary: Numerical simulation of multiphase flow plays a vital role in geoscience applications. The U-FNO neural network architecture, based on the Fourier neural operator (FNO), offers a superior and efficient solution for solving multiphase flow problems. It outperforms traditional simulators in accuracy, speed, and data utilization.
ADVANCES IN WATER RESOURCES
(2022)
Article
Environmental Studies
Abdul Motin Howlader, Dilip Patel, Robert Gammariello
Summary: This paper presents the use of an integrated deep neural network method (NNM) to predict instantaneous vehicle emissions, including CO2, CO, NOx, and HC from light-duty vehicles. Various deep-learning algorithms, such as LSTM, GRU, and RNN, were used and an integrated method combining LSTM, RNN, and GRU was employed to improve the prediction performance of vehicle emissions. The performance of each model was evaluated using MSE, RMSE, and nRMSE values. The results show that the integrated LSTM NNM outperformed the other methods in predicting vehicle emissions. This integrated model can contribute to the development of new policies and regulations for controlling vehicle emissions.
TRANSPORTATION RESEARCH PART D-TRANSPORT AND ENVIRONMENT
(2023)
Article
Environmental Sciences
Jiayi Peng, Zhenzhong Shen, Wenbing Zhang, Wen Song
Summary: The permeability characteristics of coarse-grained soil are crucial for understanding its seepage behavior and managing it effectively, which directly impacts the design, construction, and safety of embankment dams. In this study, a Convolutional Neural Network (CNN) model is used to accurately segment CT images of coarse-grained soil, surpassing traditional methods and opening new avenues for soil granulometric analysis. The reconstructed three-dimensional models from the segmented images demonstrate the effectiveness of the CNN model, highlighting its potential for automated and precise soil-particle analysis.
Article
Chemistry, Analytical
Zabir Mohammad, Arif Reza Anwary, Muhammad Firoz Mridha, Md Sakib Hossain Shovon, Michael Vassallo
Summary: A concept for a wearable monitoring framework was proposed to anticipate falls, provide safety measures, and issue notifications. The system utilized deep learning models for offline analysis and achieved high accuracy rates on the SisFall dataset.
Article
Biotechnology & Applied Microbiology
Yiqing Lan, Nannan Huang, Yiru Fu, Kehao Liu, He Zhang, Yuzhou Li, Sheng Yang
Summary: This study developed an automatic deep learning algorithm for measuring the osteogenic differentiation of stem cells accurately. The algorithm captured images of early differentiation using laser confocal scanning microscopy and successfully distinguished osteogenic cells. It also predicted the effects of small-molecule osteogenic drugs and cytokines and recognized the osteogenic differentiation of stem cells cultured on different material surfaces. This study provides an important foundation for next-generation tissue engineering and stem cell research.
FRONTIERS IN BIOENGINEERING AND BIOTECHNOLOGY
(2022)
Article
Engineering, Mechanical
Shafiq Ahmad, Zafar H. Khan, Salman Zeb, Muhammad Hamid
Summary: This study examined the effects of boundary layer flow and heat transport of a two-dimensional incompressible magnetohydrodynamic tangent hyperbolic fluid under slip boundary conditions and variable thermal conductivity. Non-similarity transformations were used to transform the governing equations into dimensionless form, and Maple software was employed for numerical solutions. The study found that various dimensionless parameters significantly affect the entropy generation rate, Bejan number, velocity, and temperature fields.
PROCEEDINGS OF THE INSTITUTION OF MECHANICAL ENGINEERS PART E-JOURNAL OF PROCESS MECHANICAL ENGINEERING
(2022)
Article
Energy & Fuels
Manita Kumari, Adil Sarwar, Mohd Tariq, Shafiq Ahmad, Adamali Shah Noor Mohamed, Eduardo M. G. Rodrigues
Summary: This paper presents a study on a three-phase multilevel inverter that utilizes switching capacitors and a single DC power supply to produce a nine-stage, three-phase voltage output. The optimal switching angles for Selective Harmonic Elimination (SHE) are determined using a recently proposed meta-heuristic technique called symbiotic organism search (SOS). The converter performance is analyzed in the MATLAB/SIMULINK environment and validated with real-time hardware-in-the-loop (HIL) experiment results.
Article
Thermodynamics
Faiz Ali, Adil Sarwar, Farhad Ilahi Bakhsh, Shafiq Ahmad, Adam Ali Shah, Hafiz Ahmed
Summary: This study aims to introduce an optimization algorithm for extracting unknown parameters of solar cells. It is found that most methods for parameter extraction in the available literature have imprecise RMSE calculation. A new optimization algorithm, the Atomic Orbital Search, is proposed and compared with other algorithms, showing better performance in parameter extraction.
ENERGY CONVERSION AND MANAGEMENT
(2023)
Article
Green & Sustainable Science & Technology
Rimsha Asad, Saud Altaf, Shafiq Ahmad, Adamali Shah Noor Mohamed, Shamsul Huda, Sofia Iqbal
Summary: With the COVID-19 pandemic, the transition from in-person to online education has posed challenges such as unfamiliar digital tools, lack of internet access, and communication barriers. To address these issues, a flexible and generalizable model was developed using a dataset of college and university students' responses, considering a broader range of attributes. By utilizing the CatBoost algorithm, the model achieved a 96.8% degree of accuracy and proved to be an effective predictor of student performance.
Article
Computer Science, Information Systems
Saud Altaf, Muhammad Haroon, Shafiq Ahmad, Emad Abouel Nasr, Mazen Zaindin, Shamsul Huda, Zia ur Rehman
Summary: This paper presents a 3D human pose recognition framework based on ANN for learning error estimation. A laboratory-based multisensory testbed was developed to validate the concept and results. The proposed method effectively estimates a person's pose in real time using RFID transceiver-based solutions.
Article
Energy & Fuels
Safwan Mustafa, Adil Sarwar, Mohd Tariq, Shafiq Ahmad, Haitham A. Mahmoud
Summary: This article introduces a new boost inverter with a nine-level quadruple voltage boosting waveform. It addresses the issue of high voltage requirement in conventional MLI when using renewable energy sources by automatically increasing the incoming voltage. The new architecture uses only a single DC source, two switching capacitors, and eleven switches. The quality of the proposed topology has been analyzed based on component count, THD, and cost, with an achieved efficiency of 97.85%. The switching order is controlled by the Nearest Level Modulation method, and the performance of the constructed converter is evaluated using MATLAB and PLECS software.
Article
Chemistry, Analytical
Asmat Ullah, Muhammad Ismail Mohmand, Hameed Hussain, Sumaira Johar, Inayat Khan, Shafiq Ahmad, Haitham A. Mahmoud, Shamsul Huda
Summary: Customer segmentation is a hot topic, and the RFMT model has been introduced to solve the problem. This study proposes a novel approach by introducing k-means, Gaussian, and DBSCAN algorithms for segmentation. Different cluster factor analysis methods are used to determine stable and distinctive clusters. The approach includes various segmentation criteria and will help improve customer relationships and targeted marketing.
Article
Green & Sustainable Science & Technology
Rimsha Asad, Saud Altaf, Shafiq Ahmad, Haitham Mahmoud, Shamsul Huda, Sofia Iqbal
Summary: Institutions of higher learning are striving to provide high-quality education. Educational data mining enables academic institutions to extract information from student data for predictions and improving online education.
Article
Green & Sustainable Science & Technology
Zain Bashir, Muhammad Amjad, Syed Farhan Raza, Shafiq Ahmad, Mali Abdollahian, Muhammad Farooq
Summary: The brick kiln industry is a large and unregulated sector in developing countries, emitting harmful pollutants that deteriorate the environment. The adoption of induced draught zigzag kilns (IDZKs) has proven to be a more efficient and environmentally-friendly alternative to conventional brick kilns. This study compares the emissions and environmental impact of both kiln types, finding that the zigzag technology outperforms the traditional kilns in all aspects, consuming 30% less energy and significantly reducing CO2 and PM2.5 emissions. The adoption of IDZKs can lead to a reduction of impact categories, such as particulate matter formation, photochemical oxidant formation, and terrestrial acidification.
Article
Green & Sustainable Science & Technology
Ghazala Kausar, Sajid Saleem, Fazli Subhan, Mazliham Mohd Suud, Mansoor Alam, M. Irfan Uddin
Summary: This paper presents the modelling of teachers and students' perceptions regarding gender bias in text books through AI. The data was collected from 470 respondents through a questionnaire using different themes. The experimental results show that the prediction of perceptions regarding gender varies according to the theme and the performances of the AI techniques. The paper contributes significantly to the field by modelling human behavior in society through AI.
Article
Chemistry, Multidisciplinary
Syed Farhan Raza, Muhammad Amjad, Kashif Ishfaq, Shafiq Ahmad, Mali Abdollahian
Summary: The largest problem with scanning real objects is the high costs and low quality of point cloud data, which leads to increased time and costs. This research aims to improve the quality of scanning to save time, costs, and computational resources. By optimizing the factors associated with a 3D scanner, errorless digital scanned data can be obtained, which can be used in various engineering and non-engineering applications.
APPLIED SCIENCES-BASEL
(2023)
Article
Computer Science, Information Systems
Muhammad Yasir Ali, Abdullah Alsaeedi, Syed Atif Ali Shah, Wael M. S. Yafooz, Asad Waqar Malik
Summary: Smart farming is crucial for increasing crop production through technological advancements, such as accurate fertilizing, watering, and pesticide application, as well as environmental monitoring. Efficient management of agricultural networks and improved communication using machine learning algorithms play a vital role in achieving sustainability in large-scale farms. Reinforcement learning is utilized to find optimal data transmission paths in our network, resulting in reduced transmission delay compared to other tested methods.
Article
Engineering, Chemical
Injila Sajid, Ayushi Gautam, Adil Sarwar, Mohd Tariq, Hwa-Dong Liu, Shafiq Ahmad, Chang-Hua Lin, Abdelaty Edrees Sayed
Summary: This research proposes the dandelion optimizer (DO) as a solution for achieving maximum power point tracking (MPPT) in photovoltaic (PV) arrays under partial shading (PS) conditions. The effectiveness of the DO algorithm in enhancing the performance of MPPT in PV arrays, particularly in challenging partial shading conditions, is demonstrated through simulation and real-time hardware-in-the-loop (HIL) results.
Article
Green & Sustainable Science & Technology
Saud Altaf, Rimsha Asad, Shafiq Ahmad, Iftikhar Ahmed, Mali Abdollahian, Mazen Zaindin
Summary: COVID-19's rapid spread has disrupted educational initiatives worldwide, including in Pakistan. The shift to distance learning has brought challenges such as reduced access to technology and unstable internet connections for students. To evaluate the effectiveness of online education, a hybrid deep learning approach has been proposed using multiple data sources.
Article
Computer Science, Information Systems
Akhtar Munir Khan, Muhammad Asif Jan, Muhammad Sagheer, Rashida Adeeb Khanum, Muhammad Irfan Uddin, Shafiq Ahmad, Shamsul Huda
Summary: This work extends an adaptive penalty function for constrained multiobjective optimization by introducing a near feasibility threshold (NFT). The proposed variants outperform the competitors on well-known benchmark test suits, indicating the effectiveness of the modified penalty function.