检索结果-内蒙古大学图书馆

Informatik-Spektrum 2003年第2期26卷 94-102页

作者： Ney, Hermann Human Language Technology and Pattern Recognition Lehrstuhl für Informatik Rheinisch-Westfälische Technische Hochschule Aachen Ahornstraße 55 52056 Aachen E-Mail: ney@informatik.rwth-aachen.de DE

Dieser Beitrag behandlt die Rolle des statistischen Ansatzes in der maschinellen (oder automatischen) Sprachverarbeitung.

关键词：

来源：评论

学校读者我要写书评

暂无评论

HTR-Flor: A Deep Learning System for Offline Handwritten Text Recognition

HTR-Flor: A Deep Learning System for Offline Handwritten Tex...

引用

Brazilian Symposium on Computer Graphics and Image Processing (SIBGRAPI)

作者： Arthur Flor de Sousa Neto Byron Leite Dantas Bezerra Alejandro Héctor Toselli Estanislau Baptista Lima Escola Politécnica de Pernambuco Universidade de Pernambuco Recife Brazil Pattern Recognition and Human Language Technology Universitat Politècnica de València València Spain

ISBN: (数字)9781728192741

ISBN: (纸本)9781728192758

In recent years, Handwritten Text Recognition (HTR) has captured a lot of attention among the researchers of the computer vision community. Current state-of-the-art approaches for offline HTR are based on Convolutional Recurrent Neural Networks (CRNNs) excel at scene text recognition. Unfortunately, deep models such as CRNNs, Recur-rent Neural Networks (RNNs) are likely to suffer from vanishing/exploding gradient problems when processing long text images, which are commonly found in scanned documents. Besides, they usually have millions of parameters which require huge amount of data, and computational resource. Recently, a new class of neural net-work architecture, called Gated Convolutional Neural Networks (Gated-CNN), has demonstrated potentials to complement CRNN methods in modeling. Therefore, in this paper, we present a new architecture for HTR, based on Gated-CNN, with fewer parameters and fewer layers, which is able to outperform the current state-of-the-art architectures for HTR. The experiment validates that the proposed model has statistically significant recognition results, surpassing previous HTR systems by an average of 33% over five important handwritten benchmark datasets. Moreover, the proposed model is able to achieve satisfactory recognition rates even in case of few training data. Finally, its compact architecture requires less computational resources, which can be applied for real-world applications that have hardware limitations, such as robots and smartphones.

关键词： Logic gates Convolution Computer architecture Text recognition Hidden Markov models Optical imaging Computational modeling

来源：评论

学校读者我要写书评

暂无评论

Warp that smile on your face: Optimal and smooth deformations for face recognition

Warp that smile on your face: Optimal and smooth deformation...

引用

International Conference on Automatic Face and Gesture Recognition

作者： Tobias Gass Leonid Pishchulin Philippe Dreuw Hermann Ney Human Language Technology and Pattern Recognition Group RWTH Aachen University Germany Computer Vision Laboratory ETH Zurich Switzerland Computer Vision and Multimodal Computing MPI Informatics Saarbruecken Germany

In this work, we present novel warping algorithms for full 2D pixel-grid deformations for face recognition. Due to high variation in face appearance, face recognition is considered a very difficult task, especially if only a single reference image, for example a mug-shot, per face is available. Usually model-based approaches with additional training data are used to cope with several types of variation occurring in facial imaging. Image warping contrarily yields a distance measure which is invariant with regard to several types of variation. This allows for precise recognition even using only very few reference observations. Due to the computationally complex problem of optimal 2D warping, pseudo-2D warping-based approaches in the past represented strong approximations of the original problem, and were mainly successful on data with low variability or rectified images. We propose a novel 2D warping method which is globally optimal and makes no prior assumtions on the data variability besides two-dimensional smootheness constraints which both avoid local mirroring and gaps and significantly speed up the optimization. Furthermore, we show that occlusion handling is imperative to obtain smooth warpings in a variety of domains. We evaluate our novel algorithm on various well known databases, such as the AR-Face and CMU-PIE database, and provide a detailed comparison to existing warping approaches. We show that by using simple relative 2D constraints, strong local features and a kernel, which is robust w.r.t. occlusions, our computationally complex approaches outperform state-of-the-art results for recognizing faces under varying expressions, occlusions and poses. Most interestingly, we achieve higher accuracy using fewer training instances per class compared to methods learning a model of the 3D shape.

关键词： Pixel Face Face recognition Optimization Databases Approximation methods Hidden Markov models

来源：评论

学校读者我要写书评

暂无评论

RADMM: RECURRENT ADAPTIVE MIXTURE MODEL WITH APPLICATIONS TO DOMAIN ROBUST language MODELING

RADMM: RECURRENT ADAPTIVE MIXTURE MODEL WITH APPLICATIONS TO...

引用

IEEE International Conference on Acoustics, Speech and Signal Processing

作者： Kazuki Irie Shankar Kumar Michael Nirschl Hank Liao Human Language Technology and Pattern Recognition Group Computer Science Department RWTH Aachen University D-52056 Aachen Germany Google Inc. New York NY 10011 USA

ISBN: (纸本)9781538646595

We present a new architecture and a training strategy for an adaptive mixture of experts with applications to domain robust language modeling. The proposed model is designed to benefit from the scenario where the training data are available in diverse domains as is the case for YouTube speech recognition. The two core components of our model are an ensemble of parallel long short-term memory (LSTM) expert layers for each domain and another LSTM based network which generates state dependent mixture weights for combining expert LSTM states by linear interpolation. The resulting model is a recurrent adaptive mixture model (RADMM) of domain experts. We train our model on 4.4B words from YouTube speech recognition data. We report results on the YouTube speech recognition test set. Compared with a background LSTM model, we obtain up to 12% relative improvement in perplexity and an improvement in word error rate from 12.3% to 12.1% while using a lattice rescoring with strong pruning.

关键词： language modeling neural networks speech recognition mixture of experts domain adaptation modelling languages Speech recognition Neural network Professional personnel YouTube Mixture models training policy new buildings

来源：评论

学校读者我要写书评

暂无评论

Calibration of Deep Probabilistic Models with Decoupled Bayesian Neural Networks

arXiv

引用

arXiv 2019年

作者： Maroñas, Juan Paredes, Roberto Ramos, Daniel PRHLT - Pattern Recognition and Human Language Technology Research Center Universitat Politècnica de Valencia Spain AUDIAS - Audio Data Intelligence and Speech Universidad Autónoma de Madrid Spain

Deep Neural Networks (DNNs) have achieved state-of-the-art accuracy performance in many tasks. However, recent works have pointed out that the outputs provided by these models are not well-calibrated, seriously limiting their use in critical decision scenarios. In this work, we propose to use a decoupled Bayesian stage, implemented with a Bayesian Neural Network (BNN), to map the uncalibrated probabilities provided by a DNN to calibrated ones, consistently improving calibration. Our results evidence that incorporating uncertainty provides more reliable probabilistic models, a critical condition for achieving good calibration. We report a generous collection of experimental results using high-accuracy DNNs in standardized image classification benchmarks, showing the good performance, flexibility and robust behavior of our approach with respect to several state-of-the-art calibration methods. Code for reproducibility is provided. Copyright © 2019, The Authors. All rights reserved.

关键词： Calibration

来源：评论

学校读者我要写书评

暂无评论

On the Benefits of Convolutional Neural Network Combinations in Offline Handwriting Recognition

On the Benefits of Convolutional Neural Network Combinations...

引用

International Workshop on Frontiers in Handwriting Recognition

作者： Dewi Suryani Patrick Doetsch Hermann Ney King Mongkut's University of Technology North Bangkok The Sirindhorn International Thai-German Graduate School of Engineering Bangkok Thailand Human Language Technology and Pattern Recognition Computer Science Department RWTH Aachen University Aachen Germany

ISBN: (纸本)9781509009824

In this paper, we elaborate the advantages of combining two neural network methodologies, convolutional neural networks (CNN) and long short-term memory (LSTM) recurrent neural networks, with the framework of hybrid hidden Markov models (HMM) for recognizing offline handwriting text. CNNs employ shift-invariant filters to generate discriminative features within neural networks. We show that CNNs are powerful tools to extract general purpose features that even work well for unknown classes. We evaluate our system on a Chinese handwritten text database and provide a GPU-based implementation that can be used to reproduce the experiments. All experiments were conducted with RWTH OCR, an open-source system developed at our institute.

关键词： Hidden Markov models Neurons Training Handwriting recognition Biological neural networks Computer architecture Logic gates

来源：评论

学校读者我要写书评

暂无评论

Mobile music modeling, analysis and recognition

Mobile music modeling, analysis and recognition

引用

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

作者： Pavel Golik Boulos Harb Ananya Misra Michael Riley Alex Rudnick Eugene Weinstein Human Language Technology and Pattern Recognition Computer Science Department RWTH Aachen University Aachen Germany Google Inc. New York NY USA School of Informatics and Computing Indiana University Bloomington IN USA

We present an analysis of music modeling and recognition techniques in the context of mobile music matching, substantially improving on the techniques presented in [1]. We accomplish this by adapting the features specifically to this task, and by introducing new modeling techniques that enable using a corpus of noisy and channel-distorted data to improve mobile music recognition quality. We report the results of an extensive empirical investigation of the system's robustness under realistic channel effects and distortions. We show an improvement of recognition accuracy by explicit duration modeling of music phonemes and by integrating the expected noise environment into the training process. Finally, we propose the use of frame-to-phoneme alignment for high-level structure analysis of polyphonic music.

关键词： Training Accuracy Hidden Markov models Music Speech recognition USA Councils

来源：评论

学校读者我要写书评

暂无评论

Influence of text line segmentation in Handwritten Text Recognition

Influence of text line segmentation in Handwritten Text Reco...

引用

International Conference on Document Analysis and Recognition

作者： Verónica Romero Joan Andreu Sánchez Vicente Bosch Katrien Depuydt Jesse de Does Pattern Recognition and Human Language Technology Universitat Politècnica de València Camí de Vera Valencia Spain Instituut voor Nederlandse Lexicologie 2300RA Matthias de Vrieshof 2–3 Leiden Netherlands

Text line segmentation is the process by which text lines in a document image are localized and extracted. It is an important step in off-line Handwritten Text Recognition (HTR) given that the input of these systems is the line image of the text to be transcribed. A myriad of solutions to the text line segmentation problem have been proposed in the literature. Although these solutions may differ greatly on what is actually applied to perform the segmentation, they can be classified by the level of precision and detail in the final extracted lines. In this paper we study the influence and real needs of different levels of precision and detail in the segmentation solutions in a real HTR task. We test three technics of text line segmentation whose output range from a simple rectangle for each line to a perfect fitted polygon surrounding the detected lines. Experiments have been carried out with a historical collection and results show that good HTR accuracy can be obtained with simple extraction algorithms.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Automatic Semantic Segmentation of Structural Elements related to the Spinal Cord in the Lumbar Region by using Convolutional Neural Networks

Automatic Semantic Segmentation of Structural Elements relat...

引用

International Conference on pattern Recognition

作者： Jhon Jairo Saenz-Gamboa Maria de la Iglesia-Vayá Jon A. Gómez Biomedical Imaging Joint Unit Foundation for the Promotion of Health and Biomedical Research FISABIO-CIPF València Spain Pattern Recognition and Human Language Technology research center Universitat Politècnica de València València Spain

This work addresses the problem of automatically segmenting the MR images corresponding to the lumbar spine. The purpose is to detect and delimit the different structural elements like vertebrae, intervertebral discs, nerves, blood vessels, etc. This task is known as semantic segmentation. The approach proposed in this work is based on convolutional neural networks whose output is a mask where each pixel from the input image is classified into one of the possible classes. Classes were defined by radiologists and correspond to structural elements and tissues. The proposed network architectures are variants of the U-Net. Several complementary blocks were used to define the variants: spatial attention models, deep supervision and multi-kernels at input, this last block type is based on the idea of inception. Those architectures which got the best results are described in this paper, and their results are discussed. Two of the proposed architectures outperform the standard U-Net used as baseline. Those architectures which got the best results are described in this paper, and their results are discussed. Two of the proposed architectures outperform the standard U-Net used as baseline.

关键词： Image segmentation Semantics Spine Morphology Computer architecture Blood vessels Fats

来源：评论

学校读者我要写书评

暂无评论

Bridging the Native language and language Variety Identification Tasks

引用

Procedia Computer Science 2017年 112卷 1554-1561页

作者： Marc Franco-Salvador Greg Kondrak Paolo Rosso Symanto Research 90425 Nuremberg Germany Pattern Recognition and Human Language Technology (PRHLT) Research Center Universitat Politècnica de València 46022 Valencia Spain Department of Computing Science University of Alberta Edmonton AB T6G 2E8 Canada

The objective of Native language Identification is to determine the native language of the author of a text that he or she wrote in another language. By contrast, language Variety Identification aims at classifying texts representing different varieties of a single language. We postulate that both tasks may be reduced to a single objective, which is to identify the language variety of the text. We design a general approach that combines string kernels and word embeddings, which capture different characteristics of texts. The results of our experiments show that the approach achieves excellent results on both tasks, without any task-specific adaptations.

关键词： Native language Identification language Variety Identification String Kernels Word Embeddings Classifier Combination

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：