检索结果-内蒙古大学图书馆

IEEE International Conference on Image Processing

作者： Shota Orihashi Yoshihiro Yamazaki Mihiro Uchida Akihiko Takashima Ryo Masumura NTT Computer and Data Science Laboratories NTT Corporation Japan

This paper proposes a knowledge distillation method for an external bidirectional language model trained by masked language modeling to achieve high accuracy in scene text recognition. In Asian languages such as Japanese, it is necessary to perform text recognition in units of multiple words or sentences rather than individual words because words are not separated by spaces, and so high-level linguistic knowledge is needed to recognize text correctly. To enhance linguistic knowledge, several methods that use an external language model have been proposed, but these methods fail to consider future context well in performing text recognition because they revise the text candidates yielded by autoregressive text recognition models, which consider mainly past context. To overcome this deficiency, our key idea is to enhance a text recognition model by utilizing knowledge of an external bidirectional language model trained by masked language modeling, which reflects not only past but also future context. So as to actively consider future context in text recognition, our proposed method introduces a distillation loss term that makes the output probability of the text recognition model closer to that of the bidirectional language model. Experiments on Japanese scene text recognition demonstrate the effectiveness of the proposed method.

关键词：

来源：评论

学校读者我要写书评

暂无评论

OnDA-DETR: Online Domain Adaptation for Detection Transformers with Self-Training Framework

OnDA-DETR: Online Domain Adaptation for Detection Transforme...

引用

IEEE International Conference on Image Processing

作者： Satoshi Suzuki Taiga Yamane Naoki Makishima Keita Suzuki Atsushi Ando Ryo Masumura NTT Computer and Data Science Laboratories NTT Corporation Japan

This paper presents a novel method for online domain adaptation (OnDA) for DEtection TRansformer (DETR)-based object detection models called OnDA-DETR. OnDA is a domain adaptation paradigm that adapts a model trained on the source domain data to perform well on the target domain in an online manner during testing, using only the unlabeled test data from the target domain. Due to challenging and realistic problem settings, OnDA has garnered significant attention. However, OnDA methods for DETR-based models, which have demonstrated excellent performance in object detection research fields, had not been developed. OnDA-DETR is the first OnDA method specifically designed for DETR-based models. OnDA-DETR incorporates a self-training framework that generates pseudo-labels for the unlabeled target domain data. To effectively incorporate the self-training framework into DETR-based models, we leverage recall-aware pseudo-labeling and quality-aware training in OnDA-DETR. Experimental results indicate that OnDA-DETR improves the performance of the source-trained model by about 3.0 % points through OnDA.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Pruning randomly initialized neural networks with iterative randomization 21

Pruning randomly initialized neural networks with iterative ...

引用

Proceedings of the 35th International Conference on Neural Information Processing Systems

作者： Daiki Chijiwa Shin'ya Yamaguchi Yasutoshi Ida Kenji Umakoshi Tomohiro Inoue NTT Computer and Data Science Laboratories NTT Corporation NTT Social Informatics Laboratories NTT Corporation

ISBN: (纸本)9781713845393

Pruning the weights of randomly initialized neural networks plays an important role in the context of lottery ticket hypothesis. Ramanujan et al. [23] empirically showed that only pruning the weights can achieve remarkable performance instead of optimizing the weight values. However, to achieve the same level of performance as the weight optimization, the pruning approach requires more parameters in the networks before pruning and thus more memory space. To overcome this parameter inefficiency, we introduce a novel framework to prune randomly initialized neural networks with iteratively randomizing weight values (IteRand). Theoretically, we prove an approximation theorem in our framework, which indicates that the randomizing operations are provably effective to reduce the required number of the parameters. We also empirically demonstrate the parameter efficiency in multiple experiments on CIFAR-10 and ImageNet.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Acoustic XR Technology Merging Real and Virtual Sounds

引用

ntt Technical Review 2024年第6期22卷 30-34页

作者： Noguchi, Kenichi Chiba, Hironobu Kako, Tatsuya Kozuka, Shihori Kurokawa, Yoshiaki Watanabe, Yuki Nakayama, Akira Ultra-Reality Computing Group NTT Computer and Data Science Laboratories

With the spread of open-ear earphones that do not cover the ear, new listening experiences are being proposed that combines real ambient sounds with virtual sounds heard from earphones. At ntt, we call this merging of the real and virtual sounds through open-ear earphones "acoustic XR (extended reality) technology," which we are now developing with a view to actual services. In this article, we describe this technology with a focus on actual trials and touch upon future developments. © 2024 Nippon Telegraph and Telephone Corp.. All rights reserved.

关键词： Merging

来源：评论

学校读者我要写书评

暂无评论

Adversarial Finetuning with Latent Representation Constraint to Mitigate Accuracy-Robustness Tradeoff

Adversarial Finetuning with Latent Representation Constraint...

引用

International Conference on computer Vision (ICCV)

作者： Satoshi Suzuki Shin’ya Yamaguchi Shoichiro Takeda Sekitoshi Kanai Naoki Makishima Atsushi Ando Ryo Masumura NTT Computer and Data Science Laboratories Kyoto University NTT Human Informatics Laboratories

This paper addresses the tradeoff between standard accuracy on clean examples and robustness against adversarial examples in deep neural networks (DNNs). Although adversarial training (AT) improves robustness, it degrades the standard accuracy, thus yielding the tradeoff. To mitigate this tradeoff, we propose a novel AT method called ARREST, which comprises three components: (i) adversarial finetuning (AFT), (ii) representation-guided knowledge distillation (RGKD), and (iii) noisy replay (NR). AFT trains a DNN on adversarial examples by initializing its parameters with a DNN that is standardly pretrained on clean examples. RGKD and NR respectively entail a regularization term and an algorithm to preserve latent representations of clean examples during AFT. RGKD penalizes the distance between the representations of the standardly pretrained and AFT DNNs. NR switches input adversarial examples to nonadversarial ones when the representation changes significantly during AFT. By combining these components, ARREST achieves both high standard accuracy and robustness. Experimental results demonstrate that ARREST mitigates the tradeoff more effectively than previous AT-based methods do.

关键词：

来源：评论

学校读者我要写书评

暂无评论

A Security Verification Framework of Cryptographic Protocols Using Machine Learning

arXiv

引用

arXiv 2023年

作者： Ohno, Kentaro Nakabayashi, Misato NTT Computer & Data Science Laboratories Japan NTT Social Informatics Laboratories Japan

We propose a security verification framework for cryptographic protocols using machine learning. In recent years, as cryptographic protocols have become more complex, research on automatic verification techniques has been focused on. The main technique is formal verification. However, the formal verification has two problems: it requires a large amount of computational time and does not guarantee decidability. We propose a method that allows security verification with computational time on the order of linear with respect to the size of the protocol using machine learning. In training machine learning models for security verification of cryptographic protocols, a sufficient amount of data, i.e., a set of protocol data with security labels, is difficult to collect from academic papers and other sources. To overcome this issue, we propose a way to create arbitrarily large datasets by automatically generating random protocols and assigning security labels to them using formal verification tools. Furthermore, to exploit structural features of protocols, we construct a neural network that processes a protocol along its series and tree structures. We evaluate the proposed method by applying it to verification of practical cryptographic protocols. © 2023, CC BY.

关键词： Machine learning

来源：评论

学校读者我要写书评

暂无评论

Covariance-Aware Feature Alignment with Pre-Computed Source Statistics for Test-Time Adaptation to Multiple Image Corruptions

Covariance-Aware Feature Alignment with Pre-Computed Source ...

引用

IEEE International Conference on Image Processing

作者： Kazuki Adachi Shin’Ya Yamaguchi Atsutoshi Kumagai NTT Computer and Data Science Laboratories Kyoto University

Real-world image recognition systems often face corrupted input images, which cause distribution shifts and degrade the performance of models. These systems often use a single prediction model in a central server and process images sent from various environments, such as cameras distributed in cities or cars. Such single models face images corrupted in heterogeneous ways in test time. Thus, they require to instantly adapt to the multiple corruptions during testing rather than being re-trained at a high cost. Test-time adaptation (TTA), which aims to adapt models without accessing the training dataset, is one of the settings that can address this problem. Existing TTA methods indeed work well on a single corruption. However, the adaptation ability is limited when multiple types of corruption occur, which is more realistic. We hypothesize this is because the distribution shift is more complicated, and the adaptation becomes more difficult in case of multiple corruptions. In fact, we experimentally found that a larger distribution gap remains after TTA. To address the distribution gap during testing, we propose a novel TTA method named Covariance-Aware Feature alignment (CAFe). We empirically show that CAFe outperforms prior TTA methods on image corruptions, including multiple types of corruptions.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Discovering Implicit Operational Knowledge by Identifying Business Processes that Involve Multiple Organizational Entities

Discovering Implicit Operational Knowledge by Identifying Bu...

引用

IEEE Annual International computer Software and Applications Conference (COMPSAC)

作者： Shinobu Saito Yukako Iimura NTT Computer and Data Science Laboratories Tokyo Japan

ISBN: (数字)9798350376968

ISBN: (纸本)9798350376975

One of the success factors of Business process outsourcing (BPO) is a comprehensive and in-depth understanding of the business processes outsourced. However, such business processes are often undocumented, and discovering them is difficult and time-consuming for BPO service providers. Previously, we introduced an approach for business process discovery that uses swim lanes to recognize that different parts of the process may be performed by different parts of an organization. It generates the business process from an event log and user information extracted from an existing system. Herein, this paper examines whether the approach aids engineers and consultants in the providers who need to investigate business processes of organizations (i.e., their customers on which they do not have sufficient knowledge) discovering the implicit operational knowledge of the target business processes (e.g., undocumented local rules). To do so, we survey an industrial workflow system from which we collected data about 2,000 events and 269 users for a two-year period of the system's operation. We conducted a study on the case by means of document evaluation and expert interviews. The study suggested that the output of the approach is a valuable process visualization for identifying knowledge that is not documented nor recognized by even experts in organizations.

关键词： Knowledge engineering Surveys data visualization Organizations Manuals Software Outsourcing

来源：评论

学校读者我要写书评

暂无评论

Coding and Debugging by Separating Secret Code Toward Secure Remote Development 23

Coding and Debugging by Separating Secret Code Toward Secure...

引用

Proceedings of the 38th IEEE/ACM International Conference on Automated Software Engineering

作者： Shinobu Saito NTT Computer and Data Science Laboratories Tokyo Japan

ISBN: (纸本)9798350329964

It is a higher priority for organizations to keep their source code secured. When a certain specific code includes a secret such as intellectual property, they need to pay special attention to prevent the secret code from leaking outside. On the other hand, sometimes code leaks comes from acts by inside programmers. This industrial paper proposes a MORDEn (Micro Organized Remote Development Environment) toward preventing code leaks. MORDEn enables programmers capable of coding and debugging by physically separating secret code from their client. We also introduce a showcase that demonstrates the feasibility of MORDEn from a case study project using it.

关键词： remote development environment

来源：评论

学校读者我要写书评

暂无评论

Towards N-version Quantum Software Systems for Reliable Classical-Quantum Computing

Towards N-version Quantum Software Systems for Reliable Clas...

引用

IEEE International Conference on Software Reliability Engineering Workshops (ISSRE Wksp)

作者： Shinobu Saito Suguru Endo Yasunari Suzuki NTT Computer and Data Science Laboratories Tokyo Japan

ISBN: (数字)9798350367041

ISBN: (纸本)9798350367058

While quantum computers have attracted much attention, dealing with computational errors due to noise effects caused by the interaction between quantum hardware and the external environment is a significant challenge. In this paper, we propose an approach to apply N-version programming (NVP) to quantum software to improve the reliability of the entire quantum software system. First, we define architecture patterns for N-version quantum software systems (NVQS) based on a combination of quantum libraries and devices. Next, given that the output of the quantum software system is a probability distribution, we came up with an evaluation strategy inspired by the concept of NVP for clustering and selecting multiple probability distributions output from NVQS. Finally, we outline our future experimental plan.

关键词： Quantum computing Conferences Noise Programming Software systems Reliability engineering Probability distribution Libraries Hardware Software reliability

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：