Article
Biochemical Research Methods
Yingxi Yang, Hui Wang, Wen Li, Xiaobo Wang, Shizhao Wei, Yulong Liu, Yan Xu
Summary: The MultiLyGAN machine learning pipeline showed good predictive performance for identifying lysine modified sites in proteins, with CWGAN being effective in addressing data imbalance. CKSAAP, PWM, and structural features were identified as the most important feature-encoding schemes.
BMC BIOINFORMATICS
(2021)
Article
Biochemical Research Methods
Florian Pichot, Virginie Marchand, Mark Helm, Yuri Motorin
Summary: The article discusses the importance of analyzing RNA modifications through deep sequencing methods, describes common analytical approaches and issues, and explores the feasibility of using machine learning algorithms to establish predictive models.
Article
Multidisciplinary Sciences
Ali Ghanbari Sorkhi, Jamshid Pirgazi, Vahid Ghasemi
Summary: This paper proposes a machine learning-based method for predicting lysine malonylation sites, which improves the prediction accuracy by extracting different features and selecting the most efficient ones. Simulation results show that the proposed method has acceptable performance compared to other methods.
SCIENTIFIC REPORTS
(2022)
Article
Computer Science, Artificial Intelligence
Mohsen Heidari, Mohammad Hossein Moattar, Hamidreza Ghaffari
Summary: Dropout is a mechanism to prevent overfitting and improve the generalization of deep neural networks. Random dropout randomly terminates nodes in each training step, which may decrease network accuracy. In dynamic dropout, the importance of each node is calculated, and important nodes are not dropped. However, the calculation of node importance is not consistent, and it is costly to calculate the importance in each training step.
Article
Computer Science, Artificial Intelligence
Minghui Wang, Lili Song, Yaqun Zhang, Hongli Gao, Lu Yan, Bin Yu
Summary: In this paper, a new prediction model, Malsite-Deep, is proposed for predicting protein malonylation sites. The model combines feature extraction and deep neural networks to achieve accurate predictions, and its performance is evaluated on multiple test sets.
KNOWLEDGE-BASED SYSTEMS
(2022)
Article
Multidisciplinary Sciences
Dong Jin Park, Min Woo Park, Homin Lee, Young-Jin Kim, Yeongsic Kim, Young Hoon Park
Summary: The study developed an optimized ensemble model combining deep learning and machine learning models to predict diseases using laboratory test results, achieving an F1-score of 81% and prediction accuracy of 92% for the five most common diseases. There were differences in predictive power and disease classification patterns between deep learning and ML models, with high efficiency in disease prediction achieved through feature importance analysis and confusion matrix.
SCIENTIFIC REPORTS
(2021)
Article
Engineering, Geological
Yong-Gook Lee, Sang-Jin Kim, Zeinep Achmet, Oh-Sung Kwon, Duhee Park, Luigi Di Sarno
Summary: Prediction models for site amplification are developed using two machine learning algorithms, random forest (RF) and deep neural network (DNN). By utilizing matrix data containing the response spectrum of the input ground motion and shear wave velocity profile, both machine learning models achieve exceptional accuracy in predicting both the linear and nonlinear amplifications, providing accurate estimates for the mean and standard deviation of site amplification. Among the two techniques, the DNN-based model demonstrates better performance.
SOIL DYNAMICS AND EARTHQUAKE ENGINEERING
(2023)
Review
Biochemical Research Methods
Yanling Zhang, Wei Qin, Chu Wang
Summary: Metabolic reprogramming of macrophages during immune activation leads to the production of diverse small molecule metabolites that induce post-translational modifications on proteins. Recent advances in chemical proteomics have enabled precise identification of these modifications, facilitating the understanding of their functional implications.
CURRENT OPINION IN BIOTECHNOLOGY
(2021)
Article
Engineering, Biomedical
S. Rajeashwari, K. Arunesh
Summary: Predicting chronic diseases is essential for saving lives and improving well-being. Traditional clinical procedures are time-consuming, so researchers have explored the use of data mining algorithms to improve accuracy. This study proposes a dual Deep CNN for feature extraction and utilizes Modified Extreme-Random Forest for classification, resulting in a system that proves to be effective.
BIOMEDICAL SIGNAL PROCESSING AND CONTROL
(2024)
Article
Green & Sustainable Science & Technology
Zhi Wang, Xianyong Peng, Shengxian Cao, Huaichun Zhou, Siyuan Fan, Kuangyu Li, Wenbo Huang
Summary: The study focuses on the selective catalytic reduction (SCR) denitrification efficiency of coal-fired boilers. Accurate predictions of NOx emissions at the SCR inlet can improve denitrification efficiency. Deep learning, random forest (RF) algorithm, and lightweight convolutional neural network (CNN) were used to develop a prediction approach. The experimental results showed that the proposed method reduced model complexity while ensuring prediction accuracy, making it suitable for online optimization of industrial pollutant control and cleaner production.
JOURNAL OF CLEANER PRODUCTION
(2023)
Article
Materials Science, Multidisciplinary
Yi Wu, Xueqin Chen, Dongqi Jiang
Summary: In this study, a deep forest (DF) model was developed to predict central deflection. It showed stronger learning and generalization abilities compared to other deep learning algorithms. The DF model with custom backend settings outperformed the highly encapsulated model with sklearn as the backend in terms of predictive performance and computation time.
Article
Engineering, Chemical
Amin Mahdavi-Meymand, Wojciech Sulisz
Summary: In this study, ARIA models were developed to enhance the prediction of boiling point rise in a multi-stage flash desalination system. The ARIA models showed greater accuracy and increased prediction efficiency compared to regular models. The ARIA-ANFIS model performed the best, reducing the error in RF predictions by 69.66%.
Article
Chemistry, Multidisciplinary
Devasahayam Arokia Balaya Rex, Arun H. Patil, Prashant Kumar Modi, Mrudula Kinarulla Kandiyil, Sandeep Kasaragod, Sneha M. Pinto, Nandita Tanneru, Puran Singh Sijwali, Thottethodi Subrahmanya Keshava Prasad
Summary: This study analyzed the proteome and post-translational modifications of P. yoelii, identifying 3124 proteins, several novel genes, and discovering the importance of mitochondrial metabolic pathways for parasite survival and drug resistance.
Article
Environmental Sciences
Maximilian Lange, Hannes Feilhauer, Ingolf Kuehn, Daniel Doktor
Summary: Information on grassland land-use intensity is crucial for understanding biodiversity, ecosystem functioning, earth system science, and environmental monitoring. However, there is a lack of large extent, high resolution data on grassland LUI. This study developed a methodology using Convolutional Neural Networks (CNN) and Sentinel-2 satellite data to map grassland LUI at high resolution and large extent.
REMOTE SENSING OF ENVIRONMENT
(2022)
Article
Biochemical Research Methods
Chunyan Ao, Quan Zou, Liang Yu
Summary: This study developed a predictor based on machine learning to identify 2'-O-methylation modification sites in RNA. The predictor showed high efficiency and accuracy in identifying modification sites across multiple species, outperforming existing tools.
BRIEFINGS IN BIOINFORMATICS
(2022)
Article
Computer Science, Software Engineering
Mridul Ghosh, Sayan Saha Roy, Himadri Mukherjee, Sk Md Obaidullah, K. C. Santosh, Kaushik Roy
Summary: Graphic-rich texts in posters, especially in movie posters, play a vital role in conveying information and genre sentiments. Recognizing and localizing these texts require specific techniques. This paper introduces a transfer learning-based approach that achieved high accuracy on a newly developed dataset, outperforming previous tools relying on handcrafted features.
Article
Computer Science, Information Systems
Ali Alwehaibi, Marwan Bikdash, Mohammad Albogmi, Kaushik Roy
Summary: This paper proposes an optimized sentiment classification method based on deep learning for dialectal Arabic short text at the document level. The research results show significant performance improvement in Arabic text classification.
JOURNAL OF KING SAUD UNIVERSITY-COMPUTER AND INFORMATION SCIENCES
(2022)
Article
Computer Science, Information Systems
Ramit Kumar Roy, Himadri Mukherjee, Kaushik Roy, Umapada Pal
Summary: Accurately recognizing destination city names is crucial for postal documents to reach their intended addresses. In India, people often mix up scripts when writing addresses due to the country's multilingual and multi script nature. This paper presents a Convolutional Neural Network (CNN) based approach for recognizing handwritten multilingual multiscript Indian city names. The proposed scheme achieves high accuracy in both single script and multi script scenarios, with a maximum accuracy of 98.01%.
MULTIMEDIA TOOLS AND APPLICATIONS
(2022)
Article
Computer Science, Artificial Intelligence
Kha Gia Quach, Ngan Le, Chi Nhan Duong, Ibsa Jalata, Kaushik Roy, Khoa Luu
Summary: Group-level emotion recognition is a growing research area that is becoming increasingly important for assessing crowds of all sizes in the security and social media domains. This work extends previous research on group-level emotion recognition from single images or videos to fully investigate expression recognition in crowd videos through an effective deep feature level fusion mechanism.
PATTERN RECOGNITION
(2022)
Article
Business
Sahana Das, Sk Md Obaidullah, Kaushik Roy, Chanchal Kumar Saha
Summary: Cardiotocography (CTG) is a widely used technique to monitor fetal health. This study uses machine learning algorithms to accurately classify the baseline and compares the results with visual estimation by obstetricians, with FURIA algorithm achieving the highest accuracy.
INTERNATIONAL JOURNAL OF BUSINESS ANALYTICS
(2022)
Article
Computer Science, Information Systems
Payel Rakshit, Somnath Chatterjee, Chayan Halder, Shibaprasad Sen, Sk Md Obaidullah, Kaushik Roy
Summary: This paper discusses the application of popular Convolutional Neural Networks (CNNs) in Bangla handwritten character recognition and evaluates the performance of each network. The study shows the superior performance of CNN models in Bangla handwritten character recognition.
MULTIMEDIA TOOLS AND APPLICATIONS
(2023)
Review
Computer Science, Information Systems
Trishita Ghosh, Shibaprasad Sen, Sk. Md. Obaidullah, K. C. Santosh, Kaushik Roy, Umapada Pal
Summary: The easy availability and rapid use of online devices have increased the demand for online handwriting recognition. This paper discusses various machine learning and deep learning approaches for recognizing online handwritten characters, words, and texts. The advantages and challenges of online handwriting recognition are also addressed.
COMPUTER SCIENCE REVIEW
(2022)
Article
Computer Science, Artificial Intelligence
Payel Rakshit, Chayan Halder, Sk Md Obaidullah, Kaushik Roy
Summary: This paper presents a multi-script text line segmentation algorithm based on newly developed light projection, start point detection, and boundary tracking methods. The proposed approach overcomes the hindrance faced by state-of-the-art methods and achieves promising results on various public handwritten datasets.
EXPERT SYSTEMS WITH APPLICATIONS
(2023)
Article
Computer Science, Artificial Intelligence
Mridul Ghosh, Himadri Mukherjee, Sk Md Obaidullah, Xiao-Zhi Gao, Kaushik Roy
Summary: Computational perception has experienced a significant transformation from handcrafted feature-based techniques to deep learning in the field of scene text identification and recognition. Over the past decade, there have been important developments and advancements in this area. The traditional handcrafted feature-based techniques have been replaced by deep learning-based techniques, leading to a new stage in scene text identification.
ARTIFICIAL INTELLIGENCE REVIEW
(2023)
Article
Computer Science, Information Systems
Ankita Dhar, Himadri Mukherjee, Kaushik Roy, K. C. Santosh, Niladri Sekhar Dash
Summary: This article introduces a hybrid approach that combines text-based and graph-based features to showcase the effectiveness of an automatic text categorization system. The approach was applied on 14,373 Bangla articles, collected from various online news corpora covering nine categories. The experiments also include the application of the features on two popular English datasets to test the system's robustness and language independency.
JOURNAL OF INFORMATION SCIENCE
(2023)
Article
Medicine, General & Internal
Sahana Das, Himadri Mukherjee, Kaushik Roy, Chanchal Kumar Saha
Summary: Cardiotocography (CTG) is currently the only non-invasive and cost-effective tool for continuous fetal health monitoring. Automated analysis of CTG remains challenging due to the complex and dynamic patterns of fetal heart, which are poorly interpreted. In this study, a machine-learning-based model using SVM, RF, MLP, and bagging was proposed, achieving high accuracy and showing potential for integration into an automated decision support system.
Article
Computer Science, Information Systems
Debjyoti Basu, Himadri Mukherjee, Matteo Marciano, Shibaprasad Sen, Sajai Vir Singh, Sk Md Obaidullah, Kaushik Roy
Summary: This research proposes a machine learning-based approach to classify the dawn and dusk time ragas in music. Mel-frequency cepstral coefficients are used for feature extraction, and a two-stage classification technique is employed, achieving promising results.
MULTIMEDIA TOOLS AND APPLICATIONS
(2023)
Article
Computer Science, Information Systems
Ankan Bhattacharyya, Somnath Chatterjee, Shibaprasad Sen, S. K. M. D. Obaidullah, Kaushik Roy
Summary: Online handwritten word recognition is still a challenging task, especially for low-resource languages like Bangla. This study explores the use of different recurrent neural network architectures to recognize online handwritten Bangla words. The challenge lies in the variable number of strokes used to write words. The developed segmentation-free recognition module achieves high accuracy by leveraging stroke features and outperforms existing techniques.
MULTIMEDIA TOOLS AND APPLICATIONS
(2023)
Article
Computer Science, Information Systems
Somnath Chatterjee, Himadri Mukherjee, Shibaprasad Sen, Sk Md Obaidullah, Kaushik Roy
Summary: Postal documents are commonly used for official communication and online shopping. Delivery delays can occur due to various handwritten scripts, necessitating the use of postal sorting facilities. To address this problem, a Deep Learning-based system is proposed to recognize handwritten city names written in six major scripts. Experimental results show high accuracy rates in both script-dependent and independent approaches.
MULTIMEDIA TOOLS AND APPLICATIONS
(2023)
Proceedings Paper
Computer Science, Theory & Methods
Taiwo Ojo, Hongmei Chi, Janei Elliston, Kaushik Roy
Summary: The probability of retrieving sensitive information from secondhand IoT devices has increased due to advancements in flash memory storage technology. This study investigates data retrieval methods from secondhand memory cards and finds that utilizing software tools is the best way to prevent data leakage.