检索结果-内蒙古大学图书馆

Decomposition and Reorganization of Phonetic Information for Speaker Embedding Learning

IEEE/ACM Transactions on Audio Speech and Language Processing 2023年 31卷 1745-1757页

作者： Hong, Qian-Bei Wu, Chung-Hsien Wang, Hsin-Min National Cheng Kung University and Academia Sinica Graduate Program of Multimedia Systems and Intelligent Computing Tainan701 Taiwan Academia Sinica Taipei115 Taiwan National Cheng Kung University Department of Computer Science and Information Engineering Tainan701401 Taiwan Academia Sinica Institute of Information Science Taipei115 Taiwan

Speech content is closely related to the stability of speaker embeddings in speaker verification tasks. In this paper, we propose a novel architecture based on self-constraint learning (SCL) and reconstruction task (RT) to remove the influence of phonetic information on speaker embedding generation. First, SCL is used to reduce the divergence of frame-level features, which can avoid ambiguity between the resulting embeddings of the two utterances being compared. Second, RT is used to further remove phonetic information in frame-level layers, focusing on speaker-discriminative feature transformation. In our experiments, the speaker embedding models were trained on the VoxCeleb2 dataset and evaluated on the VoxCeleb1, Librispeech, SITW and VoxMovies datasets. Experimental results on VoxCeleb1 show that the proposed DROP-TDNN system reduced the EER by 7.5%, compared to the state-of-the-art ECAPA-TDNN system. Furthermore, the proposed DROP-TDNN system also outperformed the ECAPA-TDNN system in the experiments on SITW, Librispeech and VoxMovies under cross-dataset conditions. In the experiments on SITW, the proposed system reduced the EER by 3.4% compared to the ECAPA-TDNN system. In the experiments on Librispeech, the proposed system demonstrated the advantage of removing phonetic information under the clean speech condition, with a significant reduction of 25.5% in EER compared to the ECAPA-TDNN system. In the experiments on VoxMovies, the proposed system reduced the EER by up to 7.9% compared to the ECAPA-TDNN system under different pronunciation and background conditions. © 2014 IEEE.

关键词： Speech recognition

来源：评论

学校读者我要写书评

暂无评论

Enhancing chemical synthesis: a two-stage deep neural network for predicting feasible reaction conditions

引用

Journal of Cheminformatics 2024年第1期16卷 11页

作者： Chen, Lung-Yi Li, Yi-Pei Department of Chemical Engineering National Taiwan University No. 1 Sec. 4 Roosevelt Road Taipei 10617 Taiwan Taiwan International Graduate Program on Sustainable Chemical Science and Technology (TIGP-SCST) No. 128 Sec. 2 Academia Road Taipei 11529 Taiwan

In the field of chemical synthesis planning, the accurate recommendation of reaction conditions is essential for achieving successful outcomes. This work introduces an innovative deep learning approach designed to address the complex task of predicting appropriate reagents, solvents, and reaction temperatures for chemical reactions. Our proposed methodology combines a multi-label classification model with a ranking model to offer tailored reaction condition recommendations based on relevance scores derived from anticipated product yields. To tackle the challenge of limited data for unfavorable reaction contexts, we employed the technique of hard negative sampling to generate reaction conditions that might be mistakenly classified as suitable, forcing the model to refine its decision boundaries, especially in challenging cases. Our developed model excels in proposing conditions where an exact match to the recorded solvents and reagents is found within the top-10 predictions 73% of the time. It also predicts temperatures within ± 20 ^∘C of the recorded temperature in 89% of test cases. Notably, the model demonstrates its capacity to recommend multiple viable reaction conditions, with accuracy varying based on the availability of condition records associated with each reaction. What sets this model apart is its ability to suggest alternative reaction conditions beyond the constraints of the dataset. This underscores its potential to inspire innovative approaches in chemical research, presenting a compelling opportunity for advancing chemical synthesis planning and elevating the field of reaction engineering. Scientific contribution: The combination of multi-label classification and ranking models provides tailored recommendations for reaction conditions based on the reaction yields. A novel approach is presented to address the issue of data scarcity in negative reaction conditions through data augmentation. Graphical Abstract: [Figure not available: see fullt

关键词： Multi-label classification Multi-task modeling Reaction condition Recommendation system

来源：评论

学校读者我要写书评

暂无评论

Catalysis-Induced Highly-Stable Interface on Porous Silicon for High-Rate Lithium-Ion Batteries

引用

Nano-Micro Letters 2025年第8期17卷 548-563页

作者： Zhuobin Han Phornphimon Maitarad Nuttapon Yodsin Baogang Zhao Haoyu Ma Kexin Liu Yongfeng Hu Siriporn Jungsuttiwong Yumei Wang Li Lu Liyi Shi Shuai Yuan Yongyao Xia Yingying Lv Research Centre of Nanoscience and Nanotechnology Shanghai UniversityShanghai 200444People’s Republic of China Department of Chemistry Faculty of ScienceSilpakorn UniversityNakhon Pathom 73000Thailand Sinopec Shanghai Research Institute of Petrochemical Technology Co. Ltd.Shanghai 201208People’s Republic of China Department of Chemistry and Center of Excellence for Innovation in Chemistry Faculty of Science Ubon Ratchathani UniversityUbon Ratchathani 34190Thailand National University of Singapore(Chongqing)Research Institute Chongqing 401123People’s Republic of China Department of Chemistry Fudan UniversityShanghai 200433People’s Republic of China Emerging Industries Institute Shanghai University Jiaxing 314006ZhejiangPeople’s Republic of China Program in Bioinformatics and Computational Biology Graduate SchoolChulalongkorn UniversityBangkok 10330Thailand

Silicon stands as a key anode material in lithium-ion battery ascribing to its high energy ***,the poor rate performance and limited cycling life remain unresolved through conventional approaches that involve carbon composites or nanostructures,primarily due to the un-controllable effects arising from the substantial formation of a solid electrolyte interphase(SEI)during the ***,an ultra-thin and homogeneous Ti doping alumina oxide catalytic interface is meticulously applied on the porous Si through a synergistic etching and hydrolysis *** defect-rich oxide interface promotes a selective adsorption of fluoroethylene carbonate,leading to a catalytic reaction that can be aptly described as“molecular concentration-in situ conversion”.The resultant inorganic-rich SEI layer is electrochemical stable and favors ion-transport,particularly at high-rate cycling and high *** robustly shielded porous Si,with a large surface area,achieves a high initial Coulombic efficiency of 84.7%and delivers exceptional high-rate performance at 25 A g^(−1)(692 mAh g^(−1))and a high Coulombic efficiency of 99.7%over 1000 *** robust SEI constructed through a precious catalytic layer promises significant advantages for the fast development of silicon-based anode in fast-charging batteries.

关键词： Catalytic interface Mesoporous Inorganic-rich SEI Silicon anode Lithium-ion batteries

来源：评论

学校读者我要写书评

暂无评论

Strategies to Reduce Long-Term Drug Resistance by Considering Effects of Differential Selective Treatments 3rd

Strategies to Reduce Long-Term Drug Resistance by Considerin...

引用

3rd International Symposium on Mathematical and computational Oncology, ISMCO 2021

作者： Ghodsi Asnaashari, Tina Chang, Young Hwan Department of Biomedical Engineering and Computational Biology Program Oregon Health and Science University Portland United States

ISBN: (纸本)9783030912406

Despite great advances in modeling and cancer therapy using optimal control theory, tumor heterogeneity and drug resistance are major obstacles in cancer treatments. Since recent biological studies demonstrated the evidence of tumor heterogeneity and assessed potential biological and clinical implications, tumor heterogeneity should be taken into account in the optimal control problem to improve treatment strategies. Here, first we study the effects of two different treatment strategies (i.e., symmetric and asymmetric) in a minimal two-population model to examine the long-term effects of these treatment methods on the system. Second, by considering tumor adaptation to treatment as a factor of the cost function, the optimal treatment strategy is derived. Numerical examples show that optimal treatment decreases tumor burden for the long-term by decreasing rate of tumor adaptation over time. © 2021, Springer Nature Switzerland AG.

关键词： Tumors

来源：评论

学校读者我要写书评

暂无评论

Improving Supplier Evaluation Model using Ensemble Method-Machine Learning for Food Industry

引用

Procedia Computer science 2023年 227卷 307-315页

作者： Muhammad Asrol Sofyan Wahyudi Suharjito Christian Harito Ditdit N Utama Muhammad Syafrudin Industrial Engineering Department BINUS Graduate Program – Master of Industrial Engineering Bina Nusantara University Jakarta 11480 Indonesia Computer Science Department BINUS Graduate Program – Master of Computer Science Bina Nusantara University Jakarta 11480 Indonesia Department of Artificial Intelligence Sejong University Seoul 05006 South Korea

Supplier evaluation has a crucial role in maintaining efficiency in the food industry supply chain. Machine learning approaches can be employed to formulate models aimed at analyzing and evaluating supplier performance. Previous research has successfully designed decision tree and neural network models for assessing suppliers in the food industry with accuracies of 84.2% and 92.8% separately. Recognizing the opportunity to improve the model's performance, this study aims to advancing the machine learning models accuracy for analyzing and evaluating suppliers in the food industry. Two main models are proposed to enhance model accuracy: ensemble methods and support vector machine. This research has successfully designed a supplier evaluation model and demonstrated that the ensemble method - gradient boosting model outperforms other ensemble methods and support vector machine which is achieved a accuracy of 93.6% on a cross-validation dataset. The development of a dashboard is required to implement the supplier evaluation model using machine learning, facilitating decision-makers in evaluating and controlling supplier performance.

关键词： Ensemble Method Food Machine Learning Supply chain Supplier evaluation

来源：评论

学校读者我要写书评

暂无评论

Accuracy of DFT computed oxygen-vacancy formation energies and high-throughput search of solar thermochemical water-splitting compounds

引用

Physical Review Materials 2023年第6期7卷 065403-065403页

作者： Bianca Baldassarri Jiangang He Xin Qian Emanuela Mastronardo Sean Griesemer Sossina M. Haile Christopher Wolverton Graduate Program in Applied Physics Northwestern University Evanston Illinois USA Department of Materials Science and Engineering Northwestern University Evanston Illinois USA

The enthalpy change involved in metal oxide reduction is a key quantity in various processes related to energy conversion and storage and is of particular interest for computational prediction. Often this prediction involves the simulation of a high-temperature reduction process with a 0-K methodology like density functional theory (DFT), and it is not infrequent for the high temperature and 0-K stable crystal structures to differ. This introduces a conundrum with regard to the choice of crystal structure to utilize in the computation, with approaches in the literature varying and experimental validation remaining scarce. In this work, we address both the crystal structure conundrum and the experimental validation, and then apply the insights we gain to guide a high-throughput search for new materials for solar thermochemical water-splitting applications. By computing the DFT+U oxygen-vacancy formation energy (ΔEvf) of a selection of ABO3 compounds and comparing different crystal structures for each composition, we highlight the issues that arise when the structure utilized in the computation is dynamically unstable at 0 K, namely the presence of an artificial lowering of ΔEvf, and the lack of convergence of ΔEvf with cell size. We solve these limitations by identifying and employing a suitable surrogate dynamically stable structure. We then validate the predictive power of our calculations against appositely generated experimental measurements of reduction enthalpy for a series of Hubbard U values, finding an accuracy ranging between 0.2–0.6 eV/O. In light of such conclusions, we revise and expand a previous a high-throughput DFT study on ABO3 perovskite oxides. We provide a list of candidate STCH materials, highlight trends with redox-active cation and structural distortion, and identify Mn4+, Mn3+, and Co3+ as the most promising redox-active cations.

关键词： Energy applications First-principles calculations Hydrogen energy

来源：评论

学校读者我要写书评

暂无评论

Classifying Tweet Sentiment Analysis on Airlines using Random Forest Classifier, Naïve Bayes Classifier and K-Nearest Neighbors Classifier Algorithm

Classifying Tweet Sentiment Analysis on Airlines using Rando...

引用

2023 International Conference on Artificial Intelligence Robotics, Signal and Image Processing, AIRoSIP 2023

作者： Alexander, Ivan Heryadi, Yaya Lukas Suparta, Wayan Sonata, Ilvico Bina Nusantara University Binus Graduate Program Computer Science Computer Science Departement Jakarta11480 Indonesia Universitas Katolik Indonesia Atma Jaya Cognitive Engineering Research Group Faculty of Engineering Jakarta12930 Indonesia Institut Teknologi Nasional Yogyakarta Faculty of Industrial Technology Department of Electrical Engineering Yogyakarta55281 Indonesia

ISBN: (纸本)9798350327472

Opinion has always affected businesses and individuals especially from the Public. People react through social media and spread it incompletely. The situation was then accepted as public opinion. There are three categories of opinion i.e., positive, negative, and neutral. To investigate public opinion on a particular issue of interest, sentiment analysis might be employed. By offering a solution to the text classification challenge, this experiment focuses on the perception of airline tweets among the American public. We offer a method to classify tweets as positive, negative, or neutral using three machine learning classifiers: Random Forest Classifier, Naive Bayes Classifier, and K-Nearest Neighbors Classifier. We examined the performance of three machine learning classifiers in terms of accuracy, average precision, average recall, and average f-1 score. The results demonstrate that the Random Forest Classifier, which produces 82% accuracy, 80% average precision, 82% average recall, and 80% average f1-score, performs the best among the machine learning classifiers. © 2023 IEEE.

关键词： Tweets

来源：评论

学校读者我要写书评

暂无评论

Natural antimicrobial properties of liquid smoke derived from cocoa pod shells in meatball preservation

引用

South African Journal of Chemical engineering 2023年 46卷 106-111页

作者： Desvita, Hera Faisal, M. Mahidin, Mahidin Suhendrayatna Doctoral Program Post Graduate Program Universitas Syiah Kuala Indonesia Research Center for Chemistry National Research and Innovation Agency B.J. Habibie Science and Techno Park Serpong Banten South Tangerang 15314 Indonesia Chemical Engineering Department Faculty of Engineering Universitas Syiah Kuala Indonesia Halal Research Center Universitas Syiah Kuala Indonesia

The liquid smoke, which is produced by pyrolyzing wood materials containing cellulose, hemicellulose, and lignin, can be used to preserve food. Phenolic, acetic, and carbonyl compounds found in liquid smoke can serve as a natural preservative. This study aimed to determine meatball preservation ability of liquid smoke from cocoa pod shells. Specifically, cocoa pod shells were converted to liquid smoke using slow pyrolysis at three different temperatures, 300 °C, 340 °C and 380 °C. Prior to its use, liquid smoke was purified using distillation at 190 °C. Liquid smoke at concentrations of 0.5%–3.0% was then used to preserve meatballs using an immersion method. Total volatile base-nitrogen (TVB-N), total plate count (TPC), and most probable number (MPN) analyses were used to evaluate preservation performance of liquid smoke. The statistical methods for MPN using one-way ANOVA followed by Tukey's tests. The SNI (Standar Nasional Indonesia) threshold to test whether meatballs remained fresh (TVB-N <20 mg N/g). TPC test results indicated that beef meatballs stay fresh at room temperature for 42 h or 48 h of storage using 0.5% or 1%–3% liquid smoke concentrations, respectively. Based on MPN tests showed meatballs could last for 48 h for all samples. Liquid smoke produced from cocoa pod shells can be used as a natural preservative for beef meatballs;however, liquid smoke concentrations influence the maximum preservation period for meatballs. © 2023 The Author(s)

关键词： Cocoa pod shells Liquid smoke MPN Pyrolysis Total plate count Total volatile base-nitrogen

来源：评论

学校读者我要写书评

暂无评论

Ultrahigh zT from strong electron-phonon interactions and a low-dimensional Fermi surface

引用

Energy and Environmental science 2024年第5期17卷 1904-1915页

作者： Ranganayakulu, V.K. Wang, Te-Hsien Chen, Cheng-Lung Huang, Angus Ma, Ma-Hsuan Wu, Chun-Min Tsai, Wei-Han Hung, Tsu-Lien Ou, Min-Nan Jeng, Horng-Tay Lee, Chih-Hao Chen, Kuei-Hsien Li, Wen-Hsien Brod, Madison K. Snyder, G. Jeffrey Chen, Yang-Yuan Institute of Physics Academia Sinica Taipei11529 Taiwan Department of Engineering and System Science National Tsing Hua University Hsinchu30013 Taiwan Taiwan International Graduate Program Taipei11529 Taiwan Department of Physics National Chung Hsing University Taichung City40227 Taiwan Graduate School of Materials Science National Yunlin University of Science and Technology Yunlin64002 Taiwan Bachelor Program in Semiconductor Materials and Fabrication Ming Chi University of Technology New Taipei City243303 Taiwan Department of Physics National Tsing Hua University Hsinchu30013 Taiwan Center for Theory and Computation National Tsing Hua University Hsinchu30013 Taiwan Physics Division National Center for Theoretical Sciences Taipei10617 Taiwan Department of Physics National Central University Jhongli32001 Taiwan National Synchrotron Radiation Research Center Hsinchu Science Park Hsinchu30076 Taiwan Institute of Atomic and Molecular Sciences Academia Sinica Taipei10617 Taiwan Department of Materials Science and Engineering Northwestern University EvanstonIL60208 United States Graduate Institute of Applied Physics National Chengchi University Taipei11605 Taiwan

The outstanding thermoelectric performance of GeTe has attracted significant attention in the research community in recent years. However, many of the underlying physical mechanisms that contribute to GeTe's exceptionally high figure of merit (zT) remain not fully understood. In this study, an Sb-Bi codoped GeTe single crystal (Ge0.86Sb0.08Bi0.06)Te with an ultrahigh zT of 2.7 at 700 K and a record high device zT of 1.41 in the temperature range of 300-773 K was synthesized and investigated. The ultrahigh zT is attributed to the extremely low lattice thermal conductivity induced by strong electron-phonon (EP) interactions as revealed by the experimentally observed Kohn anomaly, through inelastic neutron scattering (INS) measurements. First-principles calculations further demonstrate that the remarkable EP interaction arises from the Fermi surface nesting featured in a one-dimensional (double-walled) topology. Our finding unravels the ultrahigh-zT mechanism in GeTe-based materials, serving as an inspiring guide toward high thermoelectric performance. © 2024 The Royal Society of Chemistry.

关键词： Neutron scattering

来源：评论

学校读者我要写书评

暂无评论

Cortical transcriptome analysis after spinal cord injury reveals the regenerative mechanism of central nervous system in CRMP2 knock-in mice

引用

Neural Regeneration Research 2021年第7期16卷 1258-1265页

作者： Ayaka Sugeno Wenhui Piao Miki Yamazaki Kiyofumi Takahashi Koji Arikawa Hiroko Matsunaga Masahito Hosokawa Daisuke Tominaga Yoshio Goshima Haruko Takeyama Toshio Ohshima Laboratory for Molecular Brain Science Department of Life Science and Medical BioscienceGraduate School of Advanced Science and EngineeringWaseda UniversityTokyoJapan Biomolecular Engineering Laboratory Department of Life Science and Medical BioscienceGraduate School of Advanced Science and EngineeringWaseda UniversityTokyoJapan Computational Bio Big-Data Open Innovation Laboratory(CBBD-OIL) National Institute of Advanced Industrial Science and Technology(AIST)TokyoJapan Research Organization for Nano and Life Innovation Waseda UniversityTokyoJapan Institute for Advanced Research of Biosystem Dynamics Waseda Research Institute for Science and EngineeringGraduate School of Advanced Science and EngineeringWaseda UniversityTokyoJapan Cellular and Molecular Biotechnology Research Institute National Institute of Advanced Industrial Science and Technology(AIST)TsukubaJapan Department of Molecular Pharmacology and Neurobiology Graduate School of MedicineYokohama City UniversityYokohamaJapan

Recent studies have shown that mutation at Ser522 causes inhibition of collapsin response mediator protein 2(CRMP2) phosphorylation and induces axon elongation and partial recovery of the lost sensorimotor function after spinal cord injury(SCI).We aimed to reveal the intracellular mechanism in axotomized neurons in the CRMP2 knock-in(CRMP2KI) mouse model by performing transcriptome analysis in mouse sensorimotor cortex using micro-dissection punching *** to that, we analyzed the structural pathophysiology in axotomized or neighboring neurons after SCI and found that somatic atrophy and dendritic spine reduction in sensorimotor cortex were suppressed in CRMP2KI *** analysis of the transcriptome has aided in the identification of four hemoglobin genes Hba-a1, Hba-a2, Hbb-bs, and Hbb-bt that are significantly upregulated in wild-type mice with concomitant upregulation of genes involved in the oxidative phosphorylation and ribosomal pathways after ***, we observed substantial upregulation in channel activity genes and downregulation of genes regulating vesicles, synaptic function, glial cell differentiation in CRMP2KI ***, the transcriptome profile of CRMP2KI mice has been discussed wherein energy metabolism and neuronal pathways were found to be differentially *** results showed that CRMP2KI mice displayed improved SCI pathophysiology not only via microtubule stabilization in neurons, but also possibly via the whole metabolic system in the central nervous system, response changes in glial cells, and *** together, we reveal new insights on SCI pathophysiology and the regenerative mechanism of central nervous system by the inhibition of CRMP2 phosphorylation at *** these experiments were performed in accordance with the guidelines of the Institutional Animal Care and Use Committee at Waseda University, Japan(2017-A027 approved on March 21, 2017;2018-A003 approved on March 25, 2018;2019-A026 approved on March

关键词： CNS regeneration cortex CRMP2 hemoglobin metabolic pathway spinal cord injury spine transcriptome

来源：评论

学校读者我要写书评

暂无评论

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案：

请选择收藏分类：

通借通还

建议与咨询 留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案： 新增检索档案 确定 取消

请选择收藏分类： 新增自定义分类 确定 取消

通借通还

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

请选择保存的检索档案：

请选择收藏分类：