检索结果-内蒙古大学图书馆

IEEE Workshop on Automatic Speech Recognition and Understanding

作者： Houjian Guo Chaoran Liu Carlos Toshinori Ishi Hiroshi Ishiguro Interactive Robot Research Team Guardian Robot Project RIKEN Japan Graduate School of Engineering Science Osaka University Japan Advanced Telecommunications Research Institute International Japan

With the development of automatic speech recognition and text-to-speech technology, high-quality voice conversion can be achieved by extracting source content information and target speaker information to reconstruct waveforms. However, current methods still require improvement in terms of inference speed. In this study, we propose a lightweight VITS-based voice conversion model that uses the HuBERT-Soft model to extract content information features. Unlike the original VITS model, we use the inverse short-time Fourier transform to replace the most computationally expensive part. Through subjective and objective experiments on synthesized speech, the proposed model is capable of natural speech generation and it is very efficient at inference time. Experimental results show that our model can generate samples at over 5000 KHz on the 3090 GPU and over 250 KHz on the i9-10900K CPU, achieving faster speed in comparison to baseline methods using the same hardware configuration.

关键词：

来源：评论

学校读者我要写书评

暂无评论

WHO IS YOUR RIGHT MIXUP PARTNER IN POSITIVE AND UNLABELED LEARNING 10

WHO IS YOUR RIGHT MIXUP PARTNER IN POSITIVE AND UNLABELED LE...

引用

10th International Conference on Learning Representations, ICLR 2022

作者： Li, Changchun Li, Ximing Feng, Lei Ouyang, Jihong College of Computer Science and Technology Jilin University China College of Computer Science Chongqing University China Imperfect Information Learning Team RIKEN Center for Advanced Intelligence Project Japan Key Laboratory of Symbolic Computation and Knowledge Engineering of Ministry of Education China

Positive and Unlabeled (PU) learning targets inducing a binary classifier from weak training datasets of positive and unlabeled instances, which arise in many real-world applications. In this paper, we propose a novel PU learning method, namely Positive and unlabeled learning with Partially Positive Mixup (P3Mix), which simultaneously benefits from data augmentation and supervision correction with a heuristic mixup technique. To be specific, we take inspiration from the decision boundary deviation phenomenon observed in our preliminary experiments, where the learned PU boundary tends to deviate from the fully supervised boundary towards the positive side. For the unlabeled instances with ambiguous predictive results, we select their mixup partners from the positive instances around the learned PU boundary, so as to transform them into augmented instances near to the boundary yet with more precise supervision. Accordingly, those augmented instances may push the learned PU boundary towards the fully supervised boundary, thereby improving the classification performance. Comprehensive experimental results demonstrate the effectiveness of the heuristic mixup technique in PU learning and show that P3Mix can consistently outperform the state-of-the-art PU learning methods. © 2022 ICLR 2022 - 10th International Conference on Learning Representationss. All rights reserved.

关键词： Heuristic methods

来源：评论

学校读者我要写书评

暂无评论

QuickVC: MANY-TO-ANY VOICE CONVERSION USING INVERSE SHORT-TIME FOURIER TRANSFORM FOR FASTER CONVERSION

arXiv

引用

arXiv 2023年

作者： Guo, Houjian Liu, Chaoran Ishi, Carlos Toshinori Ishiguro, Hiroshi Interactive Robot Research Team RIKEN Guardian Robot Project Japan Graduate School of Engineering Science Osaka University Japan Advanced Telecommunications Research Institute International Japan

With the development of automatic speech recognition (ASR) and text-to-speech (TTS) technology, high-quality voice conversion (VC) can be achieved by extracting source content information and target speaker information to reconstruct waveforms. However, current methods still require improvement in terms of inference speed. In this study, we propose a lightweight VITS-based VC model that uses the HuBERT-Soft model to extract content information features without speaker information. Through subjective and objective experiments on synthesized speech, the proposed model demonstrates competitive results in terms of naturalness and similarity. Importantly, unlike the original VITS model, we use the inverse short-time Fourier transform (iSTFT) to replace the most computationally expensive part. Experimental results show that our model can generate samples at over 5000 KHz on the 3090 GPU and over 250 KHz on the i9-10900K CPU, achieving competitive speed for the same hardware configuration.1 Copyright © 2023, The Authors. All rights reserved.

关键词： Character recognition

来源：评论

学校读者我要写书评

暂无评论

Using Joint Training Speaker Encoder With Consistency Loss to Achieve Cross-Lingual Voice Conversion and Expressive Voice Conversion

Using Joint Training Speaker Encoder With Consistency Loss t...

引用

IEEE Workshop on Automatic Speech Recognition and Understanding

Voice conversion systems have made significant advancements in terms of naturalness and similarity in common voice conversion tasks. However, their performance in more complex tasks such as cross-lingual voice conversion and expressive voice conversion remains imperfect. In this study, we propose a novel approach that combines a joint training speaker encoder and content features extracted from the cross-lingual speech recognition model Whisper to achieve high-quality cross-lingual voice conversion. Additionally, we introduce a speaker consistency loss to the joint encoder, which improves the similarity between the converted speech and the reference speech. To further explore the capabilities of the joint speaker encoder, we use the Phonetic posteriorgram as the content feature, which enables the model to effectively reproduce both the speaker characteristics and the emotional aspects of the reference speech.

关键词：

来源：评论

学校读者我要写书评

暂无评论

One positive label is sufficient: single-positive multi-label learning with label enhancement 22

One positive label is sufficient: single-positive multi-labe...

引用

Proceedings of the 36th International Conference on Neural Information Processing Systems

作者： Ning Xu Congyu Qiao Jiaqi Lv Xin Geng Min-Ling Zhang School of Computer Science and Engineering Southeast University Nanjing China RIKEN Center for Advanced Intelligence Project Tokyo Japan

ISBN: (纸本)9781713871088

Multi-label learning (MLL) learns from the examples each associated with multiple labels simultaneously, where the high cost of annotating all relevant labels for each training example is challenging for real-world applications. To cope with the challenge, we investigate single-positive multi-label learning (SPMLL) where each example is annotated with only one relevant label and show that one can successfully learn a theoretically grounded multi-label classifier on SPMLL training examples. In this paper, a novel SPMLL method named SMILE, i.e., Singlepositive MultI-label learning with Label Enhancement, is proposed. Specifically, an unbiased risk estimator is derived, which could be guaranteed to approximately converge to the optimal risk minimizer in fully supervised learning and shows that one positive label of each instance is sufficient to train a model. Then, the corresponding empirical risk estimator is established via recovering the latent soft label as a label enhancement process, where the posterior density of the latent soft labels is approximate to the variational Beta density parameterized by an inference model. Experiments on twelve corrupted MLL datasets show the effectiveness of SMILE over several existing SPMLL approaches.

关键词：

来源：评论

学校读者我要写书评

暂无评论

USING JOINT TRAINING SPEAKER ENCODER WITH CONSISTENCY LOSS TO ACHIEVE CROSS-LINGUAL VOICE CONVERSION AND EXPRESSIVE VOICE CONVERSION

arXiv

引用

arXiv 2023年

作者： Guo, Houjian Liu, Chaoran Ishi, Carlos Toshinori Ishiguro, Hiroshi Interactive Robot Research Team Guardian Robot Project RIKEN Japan Graduate School of Engineering Science Osaka University Japan Advanced Telecommunications Research Institute International Japan

Voice conversion systems have made significant advancements in terms of naturalness and similarity in common voice conversion tasks. However, their performance in more complex tasks such as cross-lingual voice conversion and expressive voice conversion remains imperfect. In this study, we propose a novel approach that combines a jointly trained speaker encoder and content features extracted from the cross-lingual speech recognition model Whisper to achieve high-quality cross-lingual voice conversion. Additionally, we introduce a speaker consistency loss to the joint encoder, which improves the similarity between the converted speech and the reference speech. To further explore the capabilities of the joint speaker encoder, we use the Phonetic posteriorgram as the content feature, which enables the model to effectively reproduce both the speaker characteristics and the emotional aspects of the reference speech. The code and pre-trained model are open-sourced . Copyright © 2023, The Authors. All rights reserved.

关键词： Signal encoding

来源：评论

学校读者我要写书评

暂无评论

Strategic Integration of HR, Organizational Management, Big Data, IoT, and AI: A Comprehensive Framework for Future-Ready Enterprises 9

Strategic Integration of HR, Organizational Management, Big ...

引用

9th International Conference on Science, Technology, engineering and Mathematics, ICONSTEM 2024

作者： Chavan, Vaibhavi Nagendra Mohammad Roshif, U. Abhijith, R. Joshi, Amit Jha, Deepak Ravensbourne University London Project Management United Kingdom Manipal Law School Manipal Academy of Higher Education Karnataka Bengaluru India Malabar College of Advanced Studies Vengara University of Calicut Department of Management Studies Kerala India Cet School of Management College of Engineering Trivandrum Apj Abdul Kalam Technological University Kerala India Ba School of Business And Finance Riga Latvia Uttar Pradesh India

ISBN: (纸本)9798350365092

This exploration paper proposes a comprehensive frame aimed at fostering unborn-ready enterprises through the strategic integration of Human coffers(HR), Organizational Management, Big Data, the Internet of Things (IoT), and Artificial Intelligence(AI). By synthesizing these critical factors, the frame seeks to optimize organizational effectiveness, enhance decision-making processes, and acclimatize proactively to evolving request dynamics. Through a methodical review of being literature and empirical substantiation, the paper delineates the interconnectedness of these rudiments and elucidates their collaborative impact on organizational performance and dexterity. likewise, it explores perpetration strategies and implicit challenges associated with espousing such an intertwined approach. This paper not only contributes to the theoretical understanding of strategic operation but also provides practical perceptivity for directors and directors seeking to navigate the complications of the contemporary business geography and place their associations for sustained success in a decreasingly digitized and competitive terrain. © 2024 IEEE.

关键词： Data Analytics

来源：评论

学校读者我要写书评

暂无评论

Rock mass joint set identification at Draa Sfar mine in Morocco through stereographic projection and K-means clustering

引用

Mediterranean Geoscience Reviews 2024年第1期6卷 49-56页

作者： Clero, Kaoutar Ed-Diny, Said Achalhi, Mohammed Cherkaoui, Mouhamed Benzakour, Intissar Soror, Tarik Rziki, Said El Fkihi, Sanaa Nadour, Mohamed Tagemouati, Hicham Bourzeix, François National School of Mines of Rabat (ENSMR) Hadj Ahmed Cherkaoui Avenue Agdal Rabat Morocco Moroccan Foundation for Advanced Science Innovation and Research (MASCIR) Mohamed Al Jazouli Street Madinat Al Irfane Rabat Morocco REMINEX Engineering R&D and Project Management BP 469 Marrakech Morocco National School of Computer Science and Systems Analysis (ENSIAS) IRDA Team Rabat IT Center ADMIR Lab Université Mohammed V Rabat Morocco

Rock mass joint set identification, crucial for geological and geotechnical studies, is often based on stereographic projection. However, the related software programs can eventually present some access limitations. Therefore, with the development of the Artificial Intelligence field and the sharp increase in Machine Learning and Deep Learning applications, this study aims to suggest an alternative solution for rock mass joint sets identification based on the K-means clustering algorithm, an unsupervised Machine Learning model that aims to structure unlabeled data into different clusters based on their patterns. The study includes a comparative analysis of joint set clustering using the K-means model and the stereographic projection. The two methods are applied to the structural data of the Draa Sfar deep underground mine in Morocco. The results show the concordance of the joint set clustering using both stereographic projection and the K-means model, highlighted by the mean dip and dip direction obtained values. This concordance confirms that K-means clustering allows identifying rock mass joint sets if access to stereographic projection automated tools is limited. However, using K-means clustering for this purpose requires applying preprocessing methods such as the elbow method or the silhouette score to identify the clusters' number previously. In conclusion, unsupervised learning can be an alternative for identifying rock mass joint sets as long as enough and suitable structural data are accessible. © 2023, The Author(s), under exclusive licence to Springer Nature Switzerland AG.

关键词： Artificial intelligence K-means clustering Rock mass joint set Stereographic projection Structural geology Unsupervised learning

来源：评论

学校读者我要写书评

暂无评论

Proposing Tap-Track Platform for Facilitating Implementation of Button-Driven Smart Services

Proposing Tap-Track Platform for Facilitating Implementation...

引用

International Seminar on Application for Technology of Information and Communication (ISemantic)

作者： Ryota Murate Tomorou Nakahashi Sinan Chen Sachio Saiki Masahide Nakamura Kiyoshi Yasuda Yuta Tuyuzaki Faculty of Engineering Kobe University Kobe Japan Graduate School of Engineering Kobe University Kobe Japan School of Data and Innovation Kochi University of Technology Kochi Japan Kobe University Tokyo Japan RIKEN Center for Advanced Intelligence Project Tokyo Japan Oyumino Central Hospital Chiba Japan

ISBN: (数字)9798350366648

ISBN: (纸本)9798350366655

In recent years, the movement toward a super smart society as a future society is being promoted as Society 5.0. In this movement, smart services [1] have become very diverse, and the burden on users has increased. Therefore, in this paper, we focus on button-driven smart services, and propose a platform " Tap Track " that makes it easy to implement button-driven smart services. In Tap Track, users can create services that run when a button [2] is pressed with no code, and users can assign their desired services to the button. The processing that can be executed when the button is pressed is playing media files, sending commands to external services, and so on. The usage history of the button can be looked back later as a life log. In the case study, we created a weather forecast confirmation service and a life log service using Tap Track and verified its operation.

关键词： Seminars Codes Tracking Weather forecasting Pressing Media Registers History Graphical user interfaces

来源：评论

学校读者我要写书评

暂无评论

SGLT2 inhibitor empagliflozin promotes revascularization in diabetic mouse hindlimb ischemia by inhibiting ferroptosis

引用

中国药理学报（英文版） 2023年第6期44卷 1161-1174页

作者： Jing-xuan Han Lai-liu Luo Yi-cheng Wang Makoto Miyagishi Vivi Kasim Shou-rong Wu Key Laboratory for Biorheological Science and Technology of Ministry of Education College of BioengineeringChongqing UniversityChongqing 400044China The 111 Project Laboratory of Biomechanics and Tissue Repair College of BioengineeringChongqing UniversityChongqing 400044China Key Laboratory for Biorheological Science and Technology of Ministry of Education College of BioengineeringChongqing UniversityChongqing 400044China The 111 Project Laboratory of Biomechanics and Tissue Repair College of BioengineeringChongqing UniversityChongqing 400044China State and Local Joint Engineering Laboratory for Vascular Implants College of BioengineeringChongqing UniversityChongqing 400044China Molecular Composite Medicine Research Group Biomedical Research InstituteNational Institute of Advanced Industrial Science and Technology(AIST)TsukubaIbaraki 305-8566Japan

Gliflozins are known as SGLT2 inhibitors,which are used to treat diabetic patients by inhibiting glucose reabsorption in kidney proximal *** studies show that gliflozins may exert other effects independent of SGLT2 *** this study we investigated their effects on skeletal muscle cell viability and paracrine function,which were crucial for promoting revascularization in diabetic hindlimb ischemia(HLI).We showed that treatment with empagliflozin(0.1-40 μM)dose-dependently increased high glucose(25 mM)-impaired viability of skeletal muscle C2C12 ***,dapagliflozin,ertugliflozin,ipragliflozin and tofogliflozin exerted similar protective effects on skeletal muscle cells cultured under the hyperglycemic *** analysis revealed an enrichment of pathways related to ferroptosis in empagliflozin-treated C2C12 *** further demonstrated that empagliflozin and other gliflozins(10 μM)restored GPX4 expression in high glucose-treated C2C12 cells,thereby suppressing ferroptosis and promoting cell ***(10 μM)also markedly enhanced the proliferation and migration of blood vessel-forming cells by promoting paracrine function of skeletal muscle C2C12 *** diabetic HLI mice,injection of empagliflozin into the gastrocnemius muscle of the left hindlimb(10mg/kg,every 3 days for 21 days)significantly enhanced revascularization and blood perfusion ***,these results reveal a novel effect of empagliflozin,a clinical hypoglycemic gliflozin drug,in inhibiting ferroptosis and enhancing skeletal muscle cell survival and paracrine function under hyperglycemic condition via restoring the expression of *** study highlights the potential of intramuscular injection of empagliflozin for treating diabetic HLI.

关键词： diabetic hindlimb ischemia empagliflozin therapeutic angiogenesis revascularization ferroptosis

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：