检索结果-内蒙古大学图书馆

International Conference on Acoustics, Speech, and Signal Processing (ICASSP)

作者： Mohammad Zeineldeen Kartik Audhkhasi Murali Karthick Baskar Bhuvana Ramabhadran Computer Science Department Human Language Technology and Pattern Recognition RWTH Aachen University Aachen Germany Google LLC New York

This work studies knowledge distillation (KD) and addresses its constraints for recurrent neural network transducer (RNN-T) models. In hard distillation, a teacher model transcribes large amounts of unlabelled speech to train a student model. Soft distillation is another popular KD method that distills the output logits of the teacher model. Due to the nature of RNN-T alignments, applying soft distillation between RNNT architectures having different posterior distributions is challenging. In addition, bad teachers having high word-error-rate (WER) reduce the efficacy of KD. We investigate how to effectively distill knowledge from variable quality ASR teachers, which has not been studied before to the best of our knowledge. We show that a sequence-level KD, full-sum distillation, outperforms other distillation methods for RNN-T models, especially for bad teachers. We also propose a variant of full-sum distillation that distills the sequence discriminative knowledge of the teacher leading to further improvement in WER. We conduct experiments on public datasets namely SpeechStew and LibriSpeech, and on in-house production data.

关键词： Knowledge engineering Training Recurrent neural networks Transducers Production Signal processing Acoustics

来源：评论

学校读者我要写书评

暂无评论

End-To-End Training of a Neural HMM with Label and Transition Probabilities

End-To-End Training of a Neural HMM with Label and Transitio...

引用

IEEE Workshop on Automatic Speech Recognition and Understanding

作者： Daniel Mann Tina Raissi Wilfried Michel Ralf Schlüter Hermann Ney AppTek GmbH Aachen Germany Machine Learning and Human Language Technology Computer Science Department RWTH Aachen University Aachen Germany

We investigate a novel modeling approach for end-to-end neural network training using hidden Markov models (HMM) where the transition probabilities between hidden states are modeled and learned explicitly. Most contemporary sequence-to-sequence models allow for from-scratch training by summing over all possible label segmentations in a given topology. In our approach there are explicit, learnable probabilities for transitions between segments as opposed to a blank label that implicitly encodes duration *** implement a GPU-based forward-backward algorithm that enables the simultaneous training of label and transition *** investigate recognition results and additionally Viterbi alignments of our models. We find that while the transition model training does not improve recognition performance, it has a positive impact on the alignment quality. The generated alignments are shown to be viable targets in state-of-the-art Viterbi trainings.

关键词：

来源：评论

学校读者我要写书评

暂无评论

The Next Frontier: Legal Technology Systems for Document Retrieval in Court Proceedings

SSRN

引用

SSRN 2024年

作者： Kew, Stephanie Yen Nee Human Rights Lawyers Association 5 Chambers 7 Savoy Court LondonWC2R 0EX United Kingdom Department of Computer Science and Computer Engineering University of Arkansas FayettevilleAR72701 United States

The advancement of legal technology systems could potentially become promising tools for document retrieval in courtroom management, offering more streamlined legal procedures and facilitating access to court records. Within this framework, this research aims to present the current development of various legal technology systems, specifically applied to retrieving information. Through the analysis of existing literature and empirical studies, the main findings demonstrate an enhanced efficiency in processing lengthy court documents when utilizing a pre-trained language model, with a notable increment in accuracy by a margin of 6.59 points. In addition, among the various classifiers, it is revealed that the Principal Component Analysis (PCA)-based linear Support Vector Machines (SVMs) attained the greatest accuracy of up to 91.30% underscoring its effectiveness in retrieval tasks, vital for making informed decision-making in court proceedings. However, these technological advances also entail some key challenges that brought about future innovations, including the integration of facial recognition into the systems for end-to-end security with secure authentication. The overall trends in this judicial field are also highlighted which helps identify the direction of legal technology so that legal professionals can stay ahead of the latest developments, adopt new cutting-edge technologies, and put into law practices to remain competent in the evolving legal profession. © 2024, The Authors. All rights reserved.

关键词： Support vector machines

来源：评论

学校读者我要写书评

暂无评论

Deterministic Medical Image Translation via High-fidelity Brownian Bridges

arXiv

引用

arXiv 2025年

作者： He, Qisheng Summerfield, Nicholas Wang, Peiyong Glide-Hurst, Carri Dong, Ming Department of Computer Science Wayne State University MI United States Department of Human Oncology University of Wisconsin-Madison WI United States Department of Medical Physics University of Wisconsin-Madison WI United States

Recent studies have shown that diffusion models produce superior synthetic images when compared to Generative Adversarial Networks (GANs). However, their outputs are often non-deterministic and lack high fidelity to the ground truth due to the inherent randomness. In this paper, we propose a novel High-fidelity Brownian bridge model (HiFi-BBrg) for deterministic medical image translations. Our model comprises two distinct yet mutually beneficial mappings: a generation mapping and a reconstruction mapping. The Brownian bridge training process is guided by the fidelity loss and adversarial training in the reconstruction mapping. This ensures that translated images can be accurately reversed to their original forms, thereby achieving consistent translations with high fidelity to the ground truth. Our extensive experiments on multiple datasets show HiFi-BBrg outperforms state-of-the-art methods in multi-modal image translation and multi-image super-resolution. © 2025, CC BY.

关键词： Generative adversarial networks

来源：评论

学校读者我要写书评

暂无评论

Scalable Subspace Clustering via Purity Kernel Tensor Learning

Scalable Subspace Clustering via Purity Kernel Tensor Learni...

引用

Integrated Circuits and Communication Systems (ICICACS), IEEE International Conference on

作者： Rachit Adhvaryu S Rajes Kannan Sanjay C. P Pravin Dange Department of Computer Science and Engineering Faculty of Engineering and Technology Parul University Gujarat India Department of Computer Science and Engineering Chennai Institute of Technology Kundrathur India Balaji Institute of Management and Human Resource Development (BIMHRD) Sri Balaji University Pune India Symbiosis Institute of Management Studies Symbiosis International (Deemed University) Pune India

ISBN: (数字)9798331508456

ISBN: (纸本)9798331508463

Subspace clustering is a fundamental task in machine learning, particularly for high-dimensional datasets where the goal is to group data points that lie in distinct subspaces. Traditional methods, such as Sparse Subspace Clustering (SSC) and Low-Rank Representation (LRR), have demonstrated good performance but often suffer from high computational costs, particularly when applied to large-scale data. To address these limitations, this paper introduces a novel technique called Purity Kernel Tensor Learning (PKTL). PKTL combines the power of tensor decomposition with a purity kernel function to improve both the efficiency and effectiveness of subspace clustering. By utilizing tensor-based representations, the method captures higher-order relationships in the data, allowing for a more robust clustering process. The purity kernel function measures the coherence within the clusters, enhancing the identification of subspaces in complex datasets. The proposed approach not only reduces computational complexity but also improves the clustering accuracy compared to existing methods. Experiments on synthetic and real-world datasets, including ORL, Yale, and MNIST, show that PKTL outperforms conventional subspace clustering algorithms, both in terms of clustering accuracy and scalability. The results demonstrate the potential of PKTL in handling large datasets while maintaining high performance, making it a promising approach for scalable subspace clustering in real-world applications.

关键词： Tensors Accuracy Scalability Noise Clustering algorithms Machine learning Robustness Computational efficiency Noise measurement Kernel

来源：评论

学校读者我要写书评

暂无评论

Enhancing and Adversarial: Improve ASR with Speaker Labels

Enhancing and Adversarial: Improve ASR with Speaker Labels

引用

International Conference on Acoustics, Speech, and Signal Processing (ICASSP)

作者： Wei Zhou Haotian Wu Jingjing Xu Mohammad Zeineldeen Christoph Lüscher Ralf Schlüter Hermann Ney Computer Science Department Human Language Technology and Pattern Recognition RWTH Aachen University Aachen Germany AppTek GmbH Aachen Germany

ASR can be improved by multi-task learning (MTL) with domain enhancing or domain adversarial training, which are two opposite objectives with the aim to increase/decrease domain variance towards domain-aware/agnostic ASR, respectively. In this work, we study how to best apply these two opposite objectives with speaker labels to improve conformer-based ASR. We also propose a novel adaptive gradient reversal layer for stable and effective adversarial training without tuning effort. Detailed analysis and experimental verification are conducted to show the optimal positions in the ASR neural network (NN) to apply speaker enhancing and adversarial training. We also explore their combination for further improvement, achieving the same performance as i-vectors plus adversarial training. Our best speaker-based MTL achieves 7% relative improvement on the Switchboard Hub5’00 set. We also investigate the effect of such speaker-based MTL w.r.t. cleaner dataset and weaker ASR NN.

关键词： Training Artificial neural networks Switches Signal processing Multitasking Acoustics Speech processing

来源：评论

学校读者我要写书评

暂无评论

Machine Methods of Tremorogram Analysis for human Functional State Estimation 3

Machine Methods of Tremorogram Analysis for Human Functional...

引用

3rd International Conference on Neural Networks and Neurotechnologies, NeuroNT 2022

作者： Safyannikov, Nikolay M. Bureneva, Olga I. Aleksanyan, Zoya A. Saint Petersburg Electrotechnical University "leti" Department of Computer Science and Engineering Saint Petersburg Russia Institute of the Human Brain Named after N. Bechtereva Russian Academy of Sciences Saint Petersburg Russia

ISBN: (纸本)9781665467766

Methods based on the assessment of motor activity by the parameters of functional or pathological tremor are used to determine the functional state of a person. The result of tremor measurements is presented in the form of tremorograms. The result of tremor measurements is presented in the form of tremorograms, they are time series obtained on the basis of tremor amplitudes. Known methods of analyzing nonstationary series can be applied to classify tremorograms by adapting them according to the peculiarities of the initial signals. In this article we propose an algorithm for analyzing tremorograms based on their spectral characteristics and involving spectrum calculation, its smoothing, and comparison of the spectrum with a template based on the DTW method. A variant of Parkinson's disease diagnostics is shown as an example of application of the proposed algorithm. © 2022 IEEE.

关键词： Time series analysis

来源：评论

学校读者我要写书评

暂无评论

Balancing exploration and exploitation phases in whale optimization algorithm: an insightful and empirical analysis

arXiv

引用

arXiv 2023年

作者： Ahmed, Aram M. Rashid, Tarik A. Hassan, Bryar A. Majidpour, Jaffer Noori, Kaniaw A. Rahman, Chnoor Maheadeen Abdalla, Mohmad Hussein Qader, Shko M. Tayfor, Noor Mohammed, Naufel B. Department of Information Technology College of Science and Technology University of Human Development Sulaimani Iraq Department of Information Technology Kurdistan Institution for Strategic Studies and Scientific Research Sulaimani46001 Iraq Department of Computer Science and Engineering School of Science and Engineering University of Kurdistan Hewler Erbil Iraq Department of Computer Science College of Science Charmo University Chamchamal Sulaimani46023 Iraq Department of Computer Science University of Raparin Sulaimani Iraq Database Technology Department Technical College of Informatics Sulaimani Polytechnic University Sulaimani Iraq Department of Computer Science Kurdistan Technical Institute Sulaimani Iraq

Agents of any metaheuristic algorithms are moving in two modes, namely exploration and exploitation. Obtaining robust results in any algorithm is strongly dependent on how to balance between these two modes. Whale optimization algorithm as a robust and well recognized metaheuristic algorithm in the literature, has proposed a novel scheme to achieve this balance. It has also shown superior results on a wide range of applications. Moreover, in the previous chapter, an equitable and fair performance evaluation of the algorithm was provided. However, to this point, only comparison of the final results is considered, which does not explain how these results are obtained. Therefore, this chapter attempts to empirically analyze the WOA algorithm in terms of the local and global search capabilities i.e. the ratio of exploration and exploitation phases. To achieve this objective, the dimension-wise diversity measurement is employed, which, at various stages of the optimization process, statistically evaluates the population's convergence and diversity. © 2023, CC BY.

关键词： Optimization

来源：评论

学校读者我要写书评

暂无评论

Development and validation of the Remote Working Benefits & Disadvantages scale

引用

Quality and Quantity 2023年第2期57卷 1159-1183页

作者： Ingusci, Emanuela Signore, Fulvio Cortese, Claudio Giovanni Molino, Monica Pasca, Paola Ciavolino, Enrico Department of History Society and Human Studies University of Salento Studium 2000 - Building 5 - Via di Valesio 24 - Lecce (LE) Office 18 Ground Floor Lecce Italy Department of Psychology University of Turin Via Verdi 10 Torino 10124 Italy Computer Science and New Technology WSB University Gdańsk 80-266 Poland

The changes that are constantly occurring in the labour sector have led organisations and companies to move towards digital transformation. This process was accelerated by the COVID-19 pandemic and conducted to a massive recourse to the practice of remote working, which in this study is understood as the term for the way of performing work outside the usual workplace and with the support of ICT. Currently, there are no flexible scales in the literature that allow measuring the benefits and disadvantages of remote working with a single instrument. Thus, the distinction between the positive and negative consequences of working remotely, substantiated by a solid literature, provides a framework for a systematical understanding of the issue. The aim of the present study is to develop and validate a scale on remote working benefits and disadvantages (RW-B&D scale). For this end, a preliminary Exploratory Factor Analysis (EFA) with 304 participants, a tailored EFA with a sample of 301 workers and a Confirmatory Factor Analysis (CFA) with 677 workers were conducted. Participants were all Italian employees who worked remotely during the period of the COVID-19 health emergency. Data were collected between October 2020 and April 2021. The psychometric robustness of the model was assessed through bootstrap validation (5000 resamples), fit indices testing and measurement of factorial invariance. The statistical analyses demonstrated the bifactorial nature of the scale, supporting the research hypothesis. The model showed good fit indices, bootstrap validation reported statistically significant saturations, good reliability indices, and convergent and discriminant validity. Measurement invariance was tested for gender and organisational sector. The results suggested that the novel scale facilitates the quantitative measurement of the benefits and disadvantages associated with remote working in empirical terms. For this reason, it could be a streamlined and psychometrically valid

关键词： Benefits Disadvantages Remote-working Scale SEM Validation

来源：评论

学校读者我要写书评

暂无评论

Exploring the Effects of Scanpath Feature Engineering for Supervised Image Classification Models

引用

Proceedings of the ACM on human-computer interaction 2023年第ETRA期7卷 1-18页

作者： Byrne, Sean Anthony Maquiling, Virmarie Reynolds, Adam Peter Frederick Polonio, Luca Castner, Nora Kasneci, Enkelejda IMT School for Advanced Studies Lucca Piazza S.Francesco 19 LU Lucca55100 Italy Human-Computer Interaction University of Töbingen Sand 14 Töbingen72076 Germany MoMiLab IMT School for Advanced Studies Lucca Piazza S.Francesco 19 LU Lucca55100 Italy Department of Economics Management and Statistics Università Degli Studi di Milano Bicocca Piazza dell Ateneo Nuovo Milan20126 Italy University of Töbingen Sand 14 Töbingen72076 Germany Human-Centered Technologies for Learning Technical University of Munich Marsstraße 20-22 Munich80335 Germany

Image classification models are becoming a popular method of analysis for scanpath classification. To implement these models, gaze data must first be reconfigured into a 2D image. However, this step gets relatively little attention in the literature as focus is mostly placed on model configuration. As standard model architectures have become more accessible to the wider eye-Tracking community, we highlight the importance of carefully choosing feature representations within scanpath images as they may heavily affect classification accuracy. To illustrate this point, we create thirteen sets of scanpath designs incorporating different eye-Tracking feature representations from data recorded during a task-based viewing experiment. We evaluate each scanpath design by passing the sets of images through a standard pre-Trained deep learning model as well as a SVM image classifier. Results from our primary experiment show an average accuracy improvement of 25 percentage points between the best-performing set and one baseline set. © 2023 ACM.

关键词： Eye tracking

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：