Article
Chemistry, Physical
Haikuo Zhang, Zhilong Wang, Jiahao Ren, Jinyun Liu, Jinjin Li
Summary: The study utilized a machine learning method to rapidly and accurately predict the binding energies of sulfur hosts with LiPS, demonstrating the potential for efficient screening and discovery of new materials to suppress the shuttle effect.
ENERGY STORAGE MATERIALS
(2021)
Article
Mechanics
Sehyeok Oh, Seungcheol Lee, Myeonggyun Son, Jooha Kim, Hyungson Ki
Summary: In this study, an AI architecture is proposed to accurately predict flow field and drone rotor thrust using high-resolution particle images. Two deep-learning models, including a generative adversarial network (GAN) and a deep convolutional neural network, were developed. The proposed architecture demonstrates high accuracy and speed compared to traditional methods, providing a new approach for measuring complex turbulent flows.
JOURNAL OF FLUID MECHANICS
(2022)
Article
Biochemical Research Methods
Fuyi Li, Xudong Guo, Peipei Jin, Jinxiang Chen, Dongxu Xiang, Jiangning Song, Lachlan J. M. Coin
Summary: Pseudouridine is a common RNA modification found in organisms, and a new computational approach called Porpoise has been proposed for accurately identifying pseudouridine sites, showing superior performance compared to existing methods and aiding in biological hypothesis formulation.
BRIEFINGS IN BIOINFORMATICS
(2021)
Article
Materials Science, Multidisciplinary
Hang Liu, Liang Xu, Zongle Ma, Zhengquan Li, Haotian Li, Ying Zhang, Bo Zhang, Ling -Ling Wang
Summary: The bandgap of semiconductor materials can be accurately predicted using machine learning methods, with an average absolute error of 0.142 eV and a coefficient of determination of 0.977 achieved through feature processing using density functional theory. The federal learning framework is employed to forecast bandgap under different experimental conditions for small-sample datasets, with bandgap errors of compound semiconductor materials at 2-10% and 2D heterojunction semiconductor materials at 5-30%.
MATERIALS TODAY COMMUNICATIONS
(2023)
Article
Electrochemistry
Peter M. Attia, Kristen A. Severson, Jeremy D. Witmer
Summary: This study develops simple, accurate, and interpretable data-driven models for battery lifetime prediction using a previously published dataset. By introducing the concept of the capacity matrix and various feature representations, the study successfully creates a number of univariate and multivariate models that achieve comparable performance to the highest-performing models previously published for this dataset.
JOURNAL OF THE ELECTROCHEMICAL SOCIETY
(2021)
Article
Computer Science, Artificial Intelligence
Victor Coscrato, Marco H. A. Inacio, Tiago Botari, Rafael Izbicki
Summary: In recent years, supervised machine learning (ML) has made impressive progress in terms of predictive power, even reaching the state of the art and super-human level in some applications. However, the adoption of ML models in real-life applications has been slower than expected. A major drawback is the lack of user trust in these models due to their black-box nature. To address this issue, the authors developed the Neural Local Smoother (NLS), a neural network architecture that achieves both accurate predictions and easy interpretability.
Article
Public, Environmental & Occupational Health
Dong Wang, Jinbo Li, Yali Sun, Xianfei Ding, Xiaojuan Zhang, Shaohua Liu, Bing Han, Haixu Wang, Xiaoguang Duan, Tongwen Sun
Summary: The study developed an artificial intelligence algorithm that can predict sepsis early in Chinese patients, using a random forest machine learning method to build a prediction model from training data set with an AUC of 0.91, sensitivity of 87%, and specificity of 89% in the validation data set. External validation studies are needed to confirm the model's applicability in diverse patient populations and clinical practices.
FRONTIERS IN PUBLIC HEALTH
(2021)
Article
Computer Science, Information Systems
Zaakki Ahamed, Maher Khemakhem, Fathy Eassa, Fawaz Alsolami, Abdullah S. Al-Malaise Al-Ghamdi
Summary: Proactive resource management in Cloud Services is important for cost effectiveness and addressing issues such as SLA violations and resource provisioning. Workload prediction using Deep Learning (DL) is popular for analyzing cloud environment data, but the quality of the training data influences the model's performance. Existing works in this domain often lack uniformity in data sources, leading to decreased efficacy of DL models. In this study, DL models are used to analyze real-world workloads from SWF, and the LSTM model exhibits the best performance. The paper also addresses the lack of literature on DL in workload prediction in cloud computing environments.
Article
Physics, Applied
Eiki Suzuki, Kiyou Shibata, Teruyasu Mizoguchi
Summary: Bonding properties can be accurately predicted using a machine learning model based on information from isolated systems before bonding, with the density of states (DOS) serving as a powerful descriptor for predicting bonding and adsorption properties.
APPLIED PHYSICS EXPRESS
(2021)
Article
Biochemical Research Methods
Bruna Moreira da Silva, David B. Ascher, Douglas E. Pires
Summary: The ability to accurately identify B-cell epitopes is crucial for vaccine design, immunodiagnostic tests, and antibody production. Existing computational methods have limited performance and lack interpretability, hindering biological insights. To overcome these limitations, we developed epitope1D, a machine learning method that accurately identifies linear B-cell epitopes and offers interpretability through novel descriptors. Our model achieved high performance on cross-validation and blind tests, outperforming state-of-the-art tools. epitope1D represents a significant advance in predictive performance and allows for biologically meaningful features to be used for interpretation.
BRIEFINGS IN BIOINFORMATICS
(2023)
Article
Biochemistry & Molecular Biology
Sian Xiao, Hao Tian, Peng Tao
Summary: Allostery is a fundamental process in regulating protein activities, and predicting allosteric sites is crucial for the discovery of allosteric drugs. We have developed a computational model using automated machine learning, which performs well in predicting allosteric sites and has been integrated with an existing system for facilitating allosteric drug discovery.
FRONTIERS IN MOLECULAR BIOSCIENCES
(2022)
Article
Genetics & Heredity
Haodong Xu, Zhongming Zhao
Summary: In this study, a large benchmark dataset was generated, consisting of 124,879 experimentally supported linear epitope-containing regions in 3567 protein clusters from over 1.3 million B cell assays. The analysis of this dataset revealed a wide range of pathogen diversity. A ten-layer deep learning framework called NetBCE was developed for linear BCE prediction.
GENOMICS PROTEOMICS & BIOINFORMATICS
(2022)
Article
Mathematics
Eduardo Diz-Mellado, Samuele Rubino, Soledad Fernandez-Garcia, Macarena Gomez-Marmol, Carlos Rivera-Gomez, Carmen Galan-Marin
Summary: This study proposes the use of machine learning techniques to fill the research gap in courtyard thermal performance modeling, showing good performance in experimental data interpretation and pattern recognition.
Article
Biochemistry & Molecular Biology
Phasit Charoenkwan, Chanin Nantasenamat, Md Mehedi Hasan, Mohammad Ali Moni, Balachandran Manavalan, Watshara Shoombuatong
Summary: A novel machine-learning meta-predictor UMPred-FRL was developed for improved umami peptide identification, combining six machine learning algorithms and seven feature encodings to achieve more accurate performance compared to baseline models.
INTERNATIONAL JOURNAL OF MOLECULAR SCIENCES
(2021)
Article
Chemistry, Applied
Jingzi Zhang, Ke Zhang, Shaomeng Xu, Yi Li, Chengquan Zhong, Mengkun Zhao, Hua-Jun Qiu, Mingyang Qin, X. -D. Xiang, Kailong Hu, Xi Lin
Summary: Discovering new superconductors through traditional trial-and-error experimental approaches is time-consuming, and the relationships between critical temperature (T-c) and material features are still unclear. The rise of machine learning technology offers opportunities to expedite inefficient exploration processes and potentially reveal new insights into these ambiguous correlations. In this study, open-source materials data, machine learning models, and data mining methods are utilized to explore the correlation between chemical features and T-c values of superconducting materials. A new model, integrating three basic algorithms, is created to improve prediction accuracy, yielding a coefficient of determination (R-2) score of 95.9% and a root mean square error (RMSE) of 6.3 K. The study estimates the average marginal contributions of material features to determine their importance in the prediction process. Results suggest that the range of thermal conductivity plays a critical role in T-c prediction among all element features. Furthermore, the integrated ML model is employed to identify twenty potential superconducting materials with T-c values exceeding 50.0 K. This study provides insights into T-c prediction to accelerate the exploration of potential high-T-c superconductors.
JOURNAL OF ENERGY CHEMISTRY
(2023)
Article
Multidisciplinary Sciences
Yan Cheng, Zhaomeng Gao, Kun Hee Ye, Hyeon Woo Park, Yonghui Zheng, Yunzhe Zheng, Jianfeng Gao, Min Hyuk Park, Jung-Hae Choi, Kan-Hao Xue, Cheol Seong Hwang, Hangbing Lyu
Summary: Atomic-resolution Cs-corrected scanning transmission electron microscopy revealed local shifting of two oxygen positions within the unit cells of a ferroelectric thin film. Reversible transition between polar and antipolar phases was induced by applying appropriate voltages. Fatigue and rejuvenation phenomena were observed.
NATURE COMMUNICATIONS
(2022)
Article
Chemistry, Physical
Namitha Anna Koshi, Dharmapura H. K. Murthy, Sudip Chakraborty, Seung-Cheol Lee, Satadeep Bhattacharjee
Summary: Strontium titanate is widely used as a promising photocatalyst due to its unique band edge alignment. Enhancing the photocatalytic activity through the control of oxygen vacancy states and doping with p-block elements like aluminum can reduce charge trapping states in SrTiO3. Calculations based on density functional theory have shown the synergistic effect of doping with aluminum and iridium in improving the photocatalytic efficiency of SrTiO3.
ACS APPLIED ENERGY MATERIALS
(2022)
Article
Materials Science, Multidisciplinary
Anupam K. Singh, Parul Devi, Ajit K. Jena, Ujjawal Modanwal, Seung-Cheol Lee, Satadeep Bhattacharjee, Boby Joseph, Sanjay Singh
Summary: Isostructural phase transition is observed in the biskyrmion host MnNiGa under pressure, accompanied by anisotropic compression behavior. The crystal structure changes with pressure while maintaining hexagonal symmetry.
PHYSICA STATUS SOLIDI-RAPID RESEARCH LETTERS
(2022)
Article
Chemistry, Physical
Kishalay Das, Bidisha Samanta, Pawan Goyal, Seung-Cheol Lee, Satadeep Bhattacharjee, Niloy Ganguly
Summary: We present a deep-learning framework, CrysXPP, for rapid and accurate prediction of electronic, magnetic, and elastic properties of various materials. The framework intelligently designs an autoencoder, CrysAE, to capture important structural and chemical properties from a large amount of crystal graph data, achieving low prediction errors. Additionally, it includes a feature selector to interpret the model's predictions.
NPJ COMPUTATIONAL MATERIALS
(2022)
Article
Physics, Applied
Swastik Sahoo, Abhinaba Sinha, Namitha Anna Koshi, Seung-Cheol Lee, Satadeep Bhattacharjee, Bhaskaran Muralidharan
Summary: The outstanding properties of graphene have paved the way for investigating other 2D-Xene materials, with silicene being the most promising due to its compatibility with current silicon fabrication technologies. Recent studies on silicene have revealed its useful electronic and mechanical properties. In this study, a theoretical model is used to investigate the piezoresistance effect of silicene in the nanoscale regime. The obtained results suggest that silicene can be used as an interconnect in flexible electronic devices and as a reference piezoresistor in strain sensors. This research will contribute to the exploration of flexible electronics applications in other 2D-Xene materials.
JOURNAL OF PHYSICS D-APPLIED PHYSICS
(2022)
Article
Chemistry, Multidisciplinary
Tae Yun Ko, Daesin Kim, Seon Joon Kim, Hyerim Kim, Arun S. Nissimagoudar, Seung-Cheol Lee, Xiaobo Lin, Peter T. Cummings, Sehyun Doo, Seongmin Park, Tufail Hassan, Taegon Oh, Ari Chae, Jihoon Lee, Yury Gogotsi, Insik In, Chong Min Koo
Summary: The article introduces a novel ligand chemistry for MXenes using alkylated 3,4-dihydroxy-L-phenylalanine (ADOPA), which can functionalize MXene surfaces under mild reaction conditions. The ADOPA ligands form strong hydrogen-bonding and pi-electron interactions with the surface terminal groups of MXenes, while the hydrophobic fluorinated alkyl tail is compatible with organic solvents. This method produces stable colloidal solutions and liquid crystals of various MXenes in organic solvents, with excellent electrical conductivity, improved oxidation stability, and processability, enabling applications in flexible electrodes and electromagnetic interference shielding.
Article
Materials Science, Multidisciplinary
Satadeep Bhattacharjee
Summary: We propose a general rule for estimating the magnetic moments of Co2-based Heusler alloys, especially when doped with late transition metals. We introduce a descriptor that can characterize both pure Co2YZ compounds and the doped ones. Our machine-learning approach is more generic than the Slater-Pauling rule since it applies to any Co2YZ Heusler compounds, regardless of whether they are half-metals or not.
JOURNAL OF MAGNETISM AND MAGNETIC MATERIALS
(2022)
Article
Chemistry, Physical
Pritam Das, Krishnamohan Thekkepat, Young-Su Lee, Seung-Cheol Lee, Satadeep Bhattacharjee
Summary: Finding a suitable material for hydrogen storage under ambient atmospheric conditions is challenging. In this study, the hydrogen storage capacity of Ti(2)AC MAX phase and its alloys were investigated using a first principles based cluster expansion approach. It was found that hydrogen adsorption is energetically more favorable on the tetrahedral site in the Ti-A layer. Ti2CuC has the highest hydrogen adsorption energy and a Cu-doped Ti2AlxCu1-xC alloy structure can store 3.66 wt% hydrogen under ambient atmospheric conditions, surpassing Ti2AlC and Ti2CuC phases.
PHYSICAL CHEMISTRY CHEMICAL PHYSICS
(2023)
Article
Chemistry, Physical
Sujana Chandrappa, Simon Joyson Galbao, P. S. Sankara Rama Krishnan, Namitha Anna Koshi, Srewashi Das, Stephen Nagaraju Myakala, Seung-Cheol Lee, Arnab Dutta, Alexey Cherevan, Satadeep Bhattacharjee, Dharmapura H. K. Murthy
Summary: In this study, p-type BTO material with visible-light absorption (λ≤600 nm) is achieved through iridium (Ir) doping. Detailed analysis using advanced spectroscopy/microscopy tools and computational electronic structure analysis provide mechanistic insights into the n- to p-type transition. This newly developed Ir-doped BTO material shows promising applications in solar fuel generation and optoelectronics.
JOURNAL OF PHYSICAL CHEMISTRY C
(2023)
Article
Chemistry, Physical
Swetarekha Ram, Gwan Hyun Choi, Albert S. Lee, Seung-Cheol Lee, Satadeep Bhattacharjee
Summary: Efficient electrocatalysts for the oxygen evolution reaction (OER) are important for efficient energy conversion and storage. In this study, the OER activities of Co single atoms (Co-SA) adsorbed on metallic MXenes were investigated. It was found that the rate-determining step in each case was the conversion of *O from *OH. The presence of oxygen vacancies decreased the OER activity in Co-SA@Ti3C2O2-δ, while it increased the OER activity in Co-SA@Mo2CO2.
JOURNAL OF PHYSICAL CHEMISTRY C
(2023)
Article
Physics, Condensed Matter
Satadeep Bhattacharjee, Seung-Cheol Lee
Summary: A new method has been proposed to analyze magnetization dynamics in spin textures under the influence of fast electron injection from topological ferromagnetic sources. The injection of these electrons generates a non-equilibrium magnetization density in the spin-texture region, resulting in the creation of spin torques through an interaction between the dipole moment and the gauge fields. These torques can exhibit both damping-like and anti-damping-like properties, similar to spin-orbit torques. Furthermore, the interaction introduces an anomalous velocity that contributes to the transverse electrical conductivity in the spin texture, resembling the topological Hall effect.
JOURNAL OF PHYSICS-CONDENSED MATTER
(2023)
Article
Physics, Condensed Matter
Krishnamohan Thekkepat, Sumanjit Das, Debi Prosad Dogra, Kapil Gupta, Seung-Cheol Lee
Summary: This paper introduces a compressive sensing-based algorithm for efficient construction of cluster expansion (CE) Hamiltonians of multicomponent alloys. The algorithm can construct sparse and physically reasonable models from a small training set, reducing the size of the training set and improving the sampling speed. The algorithm is demonstrated on four different alloy systems and successfully reproduces known ground state orderings and order-disorder transitions.
JOURNAL OF PHYSICS-CONDENSED MATTER
(2023)
Article
Physics, Multidisciplinary
Anup Kumar Mandia, Rohit Kumar, Namitha Anna Koshi, Seung-Cheol Lee, Satadeep Bhattacharjee, Bhaskaran Muralidharan
Summary: MXene is a unique class of two-dimensional compounds with exceptional optical, electrical, chemical, and mechanical properties. This study investigates the carrier transport of MXene using first principle density functional theory calculations, revealing the effects of acoustic deformation potential scattering, piezoelectric scattering, and polar optical phonon scattering mechanisms. The results provide a foundation for ab initio-based ac-transport calculations in MXene for high-frequency applications.
Article
Physics, Applied
Kun Hee Ye, In Won Yeu, Gyuseung Han, Taeyoung Jeong, Seungjae Yoon, Dohyun Kim, Cheol Seong Hwang, Jung-Hae Choi
Summary: This study used density functional theory calculations to predict phase evolution and fractions in Hf1-xZrxO2, revealing the optimal phases under different temperatures and compositions. This is of great significance for understanding and optimizing the electrical properties of HZO.
APPLIED PHYSICS REVIEWS
(2023)
Article
Chemistry, Multidisciplinary
Namitha Anna Koshi, Anup Kumar Mandia, Bhaskaran Muralidharan, Seung-Cheol Lee, Satadeep Bhattacharjee
Summary: Hall scattering factors of Sc2CF2, Sc2CO2 and Sc2C(OH)(2) were calculated using Rode's iterative approach in conjunction with calculations based on density functional theory. The study focused on the electrical transport in these MXenes, accounting for both elastic and inelastic scattering mechanisms. It was found that polar optical phonon scattering is the most significant mechanism in these materials. The observed variation in Hall factors could have significant implications for surface group identification in MXenes.