检索结果-内蒙古大学图书馆

A new training pipeline for an improved neural transducer

学校读者我要写书评

暂无评论

arXiv 2020年

作者： Zeyer, Albert Merboldt, André Schlüter, Ralf Ney, Hermann Human Language Technology and Pattern Recognition Computer Science Department RWTH Aachen University Aachen52062 Germany AppTek GmbH Aachen52062 Germany

The RNN transducer is a promising end-to-end model candidate. We compare the original training criterion with the full marginalization over all alignments, to the commonly used maximum approximation, which simplifies, improves and speeds up our training. We also generalize from the original neural network model and study more powerful models, made possible due to the maximum approximation. We further generalize the output label topology to cover RNN-T, RNA and CTC. We perform several studies among all these aspects, including a study on the effect of external alignments. We find that the transducer model generalizes much better on longer sequences than the attention model. Our final transducer model outperforms our attention model on Switchboard 300h by over 6% relative WER. Copyright © 2020, The Authors. All rights reserved.

关键词： RNA

Assessing Trustworthy AI in Times of COVID-19: Deep Learning for Predicting a Multiregional Score Conveying the Degree of Lung Compromise in COVID-19 Patients

学校读者我要写书评

暂无评论

IEEE Transactions on Technology and Society

IEEE Transactions on Technology and Society 2022年第4期3卷 272-289页

作者： Allahabadi, Himanshi Amann, Julia Balot, Isabelle Beretta, Andrea Binkley, Charles Bozenhard, Jonas Bruneault, Frederick Brusseau, James Candemir, Sema Cappellini, Luca Alessandro Chakraborty, Subrata Cherciu, Nicoleta Cociancig, Christina Coffee, Megan Ek, Irene Espinosa-Leal, Leonardo Farina, Davide Fieux-Castagnet, Genevieve Frauenfelder, Thomas Gallucci, Alessio Giuliani, Guya Golda, Adam Van Halem, Irmhild Hildt, Elisabeth Holm, Sune Kararigas, Georgios Krier, Sebastien A. Kuhne, Ulrich Lizzi, Francesca Madai, Vince I. Markus, Aniek F. Masis, Serg Mathez, Emilie Wiinblad Mureddu, Francesco Neri, Emanuele Osika, Walter Ozols, Matiss Panigutti, Cecilia Parent, Brendan Pratesi, Francesca Moreno-Sanchez, Pedro A. Sartor, Giovanni Savardi, Mattia Signoroni, Alberto Sormunen, Hanna-Maria Spezzatti, Andy Srivastava, Adarsh Stephansen, Annette F. Theng, Lau Bee Tithi, Jesmin Jahan Tuominen, Jarno Umbrello, Steven Vaccher, Filippo Vetter, Dennis Westerlund, Magnus Wurth, Renee Zicari, Roberto V. Ey Netherlands Enterprise Intelligence Department Amsterdam1083 HP Netherlands Eth Zurich Health Ethics and Policy Lab Department of Health Sciences and Technology Zürich8092 Switzerland Center for Diplomatic and Strategic Studies Postgraduate Studies in Diplomacy and International Relations Paris75015 France Pisa56124 Italy Hackensack Meridian Health Bioethics Center EdisonNJ08820 United States University of Oxford Faculty of Philosophy OxfordOX2 6GG United Kingdom Collège André- Laurendeau Philosophie Department MontrealQCH8N 2J4 Canada Université du Québec À Montréal École des Médias MontrealQCH2L 2C4 Canada Pace University Philosophy Department New YorkNY10038 United States The Ohio State University Wexner Medical Center Department of Radiology ColumbusOH43210 United States Humanitas Research Hospital Department of Radiology Milan20089 Italy Humanitas University Department of Biomedical Sciences Milan20089 Italy University of New England Faculty of Science Agriculture Business and Law ArmidaleNSW2351 Australia University of Technology Sydney Faculty of Engineering and Information Technology SydneyNSW2007 Australia Scuola Superiore Sant'Anna European Centre of Excellence on the Regulation of Robotics and Ai Pisa56127 Italy University of Bremen Group of Computer Architecture Bremen28359 Germany New York University Grossman School of Medicine Division of Infectious Diseases and Immunology Department of Medicine New YorkNY10016 United States Digital Institute Ai Research Section Stockholm16731 Sweden Arcada University of Applied Sciences Department of Business Management and Analytics Helsinki00550 Finland University of Brescia Radiological Sciences and Public Health Department of Medical and Surgical Specialties Brescia25121 Italy Sncf Reseau Sa Ethique Groupe La Plaine93418 France Institute of Diagnostic and Interventional Radiology University Hospital Zurich Zürich8091 Switzerland Eindhoven University of Tech

This article's main contributions are twofold: 1) to demonstrate how to apply the general European Union's High-Level Expert Group's (EU HLEG) guidelines for trustworthy AI in practice for the domain of healthcare and 2) to investigate the research question of what does 'trustworthy AI' mean at the time of the COVID-19 pandemic. To this end, we present the results of a post-hoc self-assessment to evaluate the trustworthiness of an AI system for predicting a multiregional score conveying the degree of lung compromise in COVID-19 patients, developed and verified by an interdisciplinary team with members from academia, public hospitals, and industry in time of pandemic. The AI system aims to help radiologists to estimate and communicate the severity of damage in a patient's lung from Chest X-rays. It has been experimentally deployed in the radiology department of the ASST Spedali Civili clinic in Brescia, Italy, since December 2020 during pandemic time. The methodology we have applied for our post-hoc assessment, called Z-Inspection®, uses sociotechnical scenarios to identify ethical, technical, and domain-specific issues in the use of the AI system in the context of the pandemic. © 2020 IEEE.

关键词： COVID-19

FULL-SUM DECODING FOR HYBRID HMM BASED SPEECH RECOGNITION USING LSTM language MODEL

学校读者我要写书评

暂无评论

arXiv 2020年

作者： Zhou, Wei Schlüter, Ralf Ney, Hermann Human Language Technology and Pattern Recognition Computer Science Department RWTH Aachen University Aachen52074 Germany AppTek GmbH Aachen52062 Germany

In hybrid HMM based speech recognition, LSTM language models have been widely applied and achieved large improvements. The theoretical capability of modeling any unlimited context suggests that no recombination should be applied in decoding. This motivates to reconsider full summation over the HMM-state sequences instead of Viterbi approximation in decoding. We explore the potential gain from more accurate probabilities in terms of decision making and apply the full-sum decoding with a modified prefix-tree search framework. The proposed full-sum decoding is evaluated on both Switchboard and Librispeech corpora. Different models using CE and sMBR training criteria are used. Additionally, both MAP and confusion network decoding as approximated variants of general Bayes decision rule are evaluated. Consistent improvements over strong baselines are achieved in almost all cases without extra cost. We also discuss tuning effort, efficiency and some limitations of full-sum decoding. Copyright © 2020, The Authors. All rights reserved.

关键词： Decoding

Treatment Fidelity in a Feasibility Trial of the Aphasia Intervention, Virtual Elaborated Semantic Feature Analysis

学校读者我要写书评

暂无评论

International journal of language & communication disorders 2025年第3期60卷 e70054页

作者： Niamh Devane Sofia Mazzoleni Nicholas Behn Jane Marshall Stephanie Wilson Katerina Hilari Department of Language and Communication Science City St George's University of London London UK. East London NHS Foundation Trust London UK. Centre for Human-Computer Interaction Design City St George's University of London London UK.

BACKGROUND AND AIMS:The reliability and validity of an intervention can be improved by checking treatment fidelity (TF). TF methods identify core components of an intervention, check their presence (or absence) and identify threats to fidelity. The Virtual Elaborated Semantic Feature Analysis (VESFA) intervention comprised individual sessions of word-finding treatment and group sessions of conversation practice. All sessions were delivered in the virtual world of EVA Park. This paper describes the TF in the VESFA trial that explored (1) if the treatment was delivered as planned, (2) which components influenced treatment adherence scores and (3) the reliability of the fidelity checklists. METHODS AND PROCEDURES:Strategies to improve fidelity were employed in the study design, the delivery of treatment, treatment receipt and treatment enactment. Two fidelity checklists were developed with input from advisors with aphasia to establish the core components of the intervention (individual and group). During the trial, treatment sessions were video-recorded. A sample of 20% of sessions was randomly selected for adherence rating. Seven research students were trained to rate the videos using the fidelity checklists. Inter- and intra-rater reliability was established. OUTCOMES AND RESULTS:Study design strategies ensured 94% of sessions ran as planned and 75% of participants (12/16) received over 90% (>36/40h) of the intended dose. The average TF across all sessions rated was 81%, demonstrating a high degree of fidelity in the delivery of the VESFA intervention. The fidelity of the individual sessions was lower (78%) than the group elements (84%). The components that most threatened treatment adherence were (1) providing a rationale for the activities and (2) specific feedback for performance. Nevertheless, participants consistently practised target words both in individual sessions and in conversations in the group sessions, demonstrating treatment receipt. Ninety-four percen

关键词： aphasia treatment validity

The rwth asr system for ted-lium release 2: improving hybrid hmm with specaugment

学校读者我要写书评

暂无评论

arXiv 2020年

作者： Zhou, Wei Michel, Wilfried Irie, Kazuki Kitza, Markus Schlüter, Ralf Ney, Hermann Human Language Technology and Pattern Recognition Computer Science Department RWTH Aachen University Aachen52074 Germany AppTek GmbH Aachen52062 Germany

We present a complete training pipeline to build a state-of-the-art hybrid HMM-based ASR system on the 2nd release of the TED-LIUM corpus. Data augmentation using SpecAugment is successfully applied to improve performance on top of our best SAT model using i-vectors. By investigating the effect of different maskings, we achieve improvements from SpecAugment on hybrid HMM models without increasing model size and training time. A subsequent sMBR training is applied to fine-tune the final acoustic model, and both LSTM and Transformer language models are trained and evaluated. Our best system achieves a 5.6% WER on the test set, which outperforms the previous state-of-the-art by 27% relative. Copyright © 2020, The Authors. All rights reserved.

关键词： Speech recognition

A Movie Recommender System Based on User Profile and Artificial Bee Colony Optimization

学校读者我要写书评

暂无评论

Computational intelligence and neuroscience 2023年第1期2023卷 2311817页

作者： Faezeh Rajabi Kouchi Sahar Oftadeh Balani Amirhossein Esmaeilpour Masoume Shafieian Rzgar Sirwan Adil Hussein Mohammed Department of Computer Engineering Central Tehran Branch Islamic Azad University Tehran Iran. Department of Computer Science Yadegar-e-Imam Khomeini (RAH) Shahre Rey Branch Islamic Azad University Tehran Iran. Computer Engineering Department Shomal University Amol Iran. IRIBU University Department of Technology and Media Engineering Tehran Iran. Department of Computer Science College of Science and Technology University of Human Development Sulaymaniyah Iraq. Department of Communication and Computer Engineering Faculty of Engineering Cihan University-Erbil Kurdistan Region Erbil Iraq.

In this study, a new algorithm for recommending movies to viewers has been proposed. To do this, the suggested method employs data mining techniques. The proposed method includes three steps for generating recommendations: "preprocessing of user profile information," "feature extraction," and "recommendation." In the first step of proposed method, the user information will be examined and transformed into a form that can be handled in the next phases. In the second step of the proposed method, user attributes are then extracted as a collection of their individual qualities, as well as the average rating of each user for various genres. The bee colony optimization algorithm is then used to select the optimal features. Finally, in the third step of the proposed method, the ratings of similar users are utilized to offer movies to the target user, and the similarities between various users are determined using the characteristics calculated for them, as well as the Euclidean distance criteria. The proposed method was evaluated using the MovieLens database, and its output was assessed in terms of precision and recall criteria; these results show that the proposed method will increase the precision by an average of 1.39% and the recall by 0.8% compared to the compared algorithms.

关键词：

A novel fuzzy trust-based secure routing scheme in flying ad hoc networks

学校读者我要写书评

暂无评论

Vehicular Communications 2023年 44卷

作者： Hosseinzadeh, Mehdi Mohammed, Adil Hussein Alenizi, Farhan A. Malik, Mazhar Hussain Yousefpoor, Efat Yousefpoor, Mohammad Sadegh Ahmed, Omed Hassan Rahmani, Amir Masoud Tightiz, Lilia Institute of Research and Development Duy Tan University Da Nang Viet Nam School of Medicine and Pharmacy Duy Tan University Da Nang Viet Nam Department of Communication and Computer Engineering Faculty of Engineering Cihan University-Erbil Kurdistan Region Iraq Electrical Engineering Department College of Engineering Prince Sattam Bin Abdulaziz University Al-Kharj 11942 Saudi Arabia School of Computing and Creative Technologies College of Arts Technology and Environment (CATE) University of the West of England Frenchay Campus Coldharbour Lane Bristol BS16 1QY United Kingdom Department of Computer Engineering Dezful Branch Islamic Azad University Dezful Iran Department of Information Technology University of Human Development Sulaymaniyah Iraq Future Technology Research Center National Yunlin University of Science and Technology Yunlin Taiwan School of Computing Gachon University 1342 Seongnamdaero Seongnam 13120 South Korea

Today, many studies assess vulnerabilities, threats, and attacks in flying ad hoc networks (FANETs) to provide solutions for countermeasures. Protecting FANETs against attackers and coordinating connections are challenging. The purpose of this study is to increase and maintain communication security. In this paper, a fuzzy trust-based secure routing scheme (FTSR) is presented in FANETs. FTSR utilizes two trust assessment mechanisms, namely local trust and path trust. Local trust strategy is a distributed process for finding reliable neighboring nodes and isolating hostile nodes on the network. In this regard, only reliable nodes are allowed to contribute to the path discovery procedure. This lowers the risk of forming fake paths in FANETs. Path trust strategy is responsible for identifying hostile nodes that are not identified in the local trust process. This strategy shows a general view of the trust status of the desired path. To design this mechanism, the source node runs a fuzzy system to select the safest path between source and the destination. Finally, network simulator 2 (NS2) implements FTSR, and the results such as malicious detection rate, packet delivery ratio, packet loss, accuracy, and delay are obtained from the simulation process. These results indicate that FTSR presents better performance compared to TOPCM, MNRiRIP, and MNDA. However, FTSR takes more time to find paths compared to TOPCM. © 2023 Elsevier Inc.

关键词： Artificial intelligence (AI) Cybersecurity Flying ad hoc networks (FANETs) Fuzzy logic Routing

-Neural network-based optimization of hydrogen fuel production energy system with proton exchange electrolyzer supported nanomaterial

学校读者我要写书评

暂无评论

Fuel 2023年第Part1期332卷

作者： Hai, Tao Hikmat Hama Aziz, Kosar Zhou, Jincheng Dhahad, Hayder A. Sharma, Kamal Fahad Almojil, Sattam Ibrahim Almohana, Abdulaziz Fahmi Alali, Abdulrhman Ismail Kh, Teeba Mehrez, Sadok Abdelrahman, Anas School of Computer and Information Qiannan Normal University for Nationalities Guizhou Duyun 558000 China Key Laboratory of Complex Systems and Intelligent Optimization of Guizhou Guizhou Duyun 558000 China nstitute for Big Data Analytics and Artificial Intelligence (IBDAAI) Universiti Teknologi MARA Selangor Shah Alam 40450 Malaysia Department of Chemistry College of Science University of Sulaimani Qlyasan Street Kurdistan Region 46001 Iraq Department of Medical Laboratory of Science College of Health Sciences University of Human Development Sulaimaniyah Iraq Mechanical Engineering Department University of Technology Baghdad Iraq Institute of Engineering and Technology GLA University Uttar Pradesh Mathura 281406 India Department of Civil Engineering College of Engineering King Saud University P.O. Box 800 Riyadh 11421 Saudi Arabia Department of Computer Engineering College of Engineering and Computer Science Lebanese French University Kurdistan Region Iraq Department of Mechanical Engineering College of Engineering at Al Kharj Prince Sattam bin Abdulaziz University 16273 Saudi Arabia Department of Mechanical Engineering University of Tunis El Manar ENIT BP 37 Le Belvédère Tunis 1002 Tunisia Department of Chemical Engineering Faculty of Engineering & Technology Future University in Egypt New Cairo 11500 Tunisia Key Laboratory of Complex Systems and Intelligent Optimization of Qiannan Duyun 558000 China

In this research, we try to investigate a solar-geothermal energy system. This system includes three turbines for power production, a PEM electrolyzer for hydrogen production, and a thermoelectric for generating electricity from excess heat. In addition, the seawater will be passed through the osmotic cycle to gain fresh water. The required power for this osmotic cycle will be obtained through the energy produced by the main turbines. The generated load, hydrogen production flow rate, purified water flow rate, and heating consumption are assessed in this study. The results showed that this system can produce 3.8 megawatts of electricity as well as 8 g per second of hydrogen fuel at the operating point. Also, the energy efficiency of this system is estimated to be 19%. Afterward, machine learning methods are used to optimize designing parameters, and the optimum operating point in terms of useful power and stored fuel flow rate is obtained by a genetic algorithm. The optimum operating point of this energy system has a useful power output of 4.099 megawatts and a hydrogen flow rate of 29 g per second. In the end, the distribution of the design parameters is displayed for points of the beam curve. © 2022

关键词： Biofuel Electrolyzer Fuel production Hydrogen fuel Optimization Solar energy

Investigation of large-margin softmax in neural language modeling

学校读者我要写书评

暂无评论

arXiv 2020年

作者： Huo, Jingjing Gao, Yingbo Wang, Weiyue Schlüter, Ralf Ney, Hermann Human Language Technology and Pattern Recognition Group Computer Science Department RWTH Aachen University Aachen52074 Germany AppTek GmbH Aachen52062 Germany

To encourage intra-class compactness and inter-class separability among trainable feature vectors, large-margin softmax methods are developed and widely applied in the face recognition community. The introduction of the large-margin concept into the softmax is reported to have good properties such as enhanced discriminative power, less overfitting and well-defined geometric intuitions. Nowadays, language modeling is commonly approached with neural networks using softmax and cross entropy. In this work, we are curious to see if introducing large-margins to neural language models would improve the perplexity and consequently word error rate in automatic speech recognition. Specifically, we first implement and test various types of conventional margins following the previous works in face recognition. To address the distribution of natural language data, we then compare different strategies for word vector norm-scaling. After that, we apply the best norm-scaling setup in combination with various margins and conduct neural language models rescoring experiments in automatic speech recognition. We find that although perplexity is slightly deteriorated, neural language models with large-margin softmax can yield word error rate similar to that of the standard softmax baseline. Finally, expected margins are analyzed through visualization of word vectors, showing that the syntactic and semantic relationships are also preserved. Copyright © 2020, The Authors. All rights reserved.

关键词： Speech recognition