检索结果-内蒙古大学图书馆

ABINet plus plus : Autonomous, Bidirectional and Iterative language modeling for Scene Text Spotting

IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE 2023年第6期45卷 7123-7141页

作者： Fang, Shancheng Mao, Zhendong Xie, Hongtao Wang, Yuxin Yan, Chenggang Zhang, Yongdong Univ Sci & Technol China Sch Informat Sci & Technol Hefei 230022 Anhui Peoples R China Hangzhou Dianzi Univ Sch Automat Hangzhou 310018 Zhejiang Peoples R China

Scene text spotting is of great importance to the computer vision community due to its wide variety of applications. Recent methods attempt to introduce linguistic knowledge for challenging recognition rather than pure visual classification. However, how to effectively model the linguistic rules in end-to-end deep networks remains a research challenge. In this paper, we argue that the limited capacity of language models comes from 1) implicit language modeling;2) unidirectional feature representation;and 3) language model with noise input. Correspondingly, we propose an autonomous, bidirectional and iterative ABINet++ for scene text spotting. First, the autonomous suggests enforcing explicitly language modeling by decoupling the recognizer into vision model and language model and blocking gradient flow between both models. Second, a novel bidirectional cloze network (BCN) as the language model is proposed based on bidirectional feature representation. Third, we propose an execution manner of iterative correction for the language model which can effectively alleviate the impact of noise input. Additionally, based on an ensemble of the iterative predictions, a self-training method is developed which can learn from unlabeled images effectively. Finally, to polish ABINet++ in long text recognition, we propose to aggregate horizontal features by embedding Transformer units inside a U-Net, and design a position and content attention module which integrates character order and content to attend to character features precisely. ABINet++ achieves state-of-the-art performance on both scene text recognition and scene text spotting benchmarks, which consistently demonstrates the superiority of our method in various environments especially on low-quality images. Besides, extensive experiments including in English and Chinese also prove that, a text spotter that incorporates our language modeling method can significantly improve its performance both in accuracy and speed compared

关键词： Text recognition Iterative methods Linguistics Transformers Character recognition Visualization Predictive models Scene text spotting scene text recognition language modeling

来源：评论

学校读者我要写书评

暂无评论

Improving speech recognition systems for the morphologically complex Malayalam language using subword tokens for language modeling

引用

EURASIP JOURNAL ON AUDIO SPEECH AND MUSIC PROCESSING 2023年第1期2023卷 47页

作者： Manohar, Kavya Jayan, A. R. Rajan, Rajeev Coll Engn Trivandrum Thiruvananthapuram 695016 Kerala India APJ Abdul Kalam Technol Univ Thiruvananthapuram Kerala India Govt Engn Coll Trichur Kerala India

This article presents the research work on improving speech recognition systems for the morphologically complex Malayalam language using subword tokens for language modeling. The speech recognition system is built using a deep neural network-hidden Markov model (DNN-HMM)-based automatic speech recognition (ASR). We propose a novel method, syllable-byte pair encoding (S-BPE), that combines linguistically informed syllable tokenization with the data-driven tokenization method of byte pair encoding (BPE). The proposed method ensures words are always segmented at valid pronunciation boundaries. On a text corpus that has been divided into tokens using the proposed method, we construct statistical n-gram language models and assess the modeling effectiveness in terms of both information-theoretic and corpus linguistic metrics. A comparative study of the proposed method with other data-driven (BPE, Morfessor, and Unigram), linguistic (Syllable), and baseline (Word) tokenization algorithms is also presented. Pronunciation lexicons of subword tokenized units are built with pronunciation described as graphemes. We develop ASR systems employing the subword tokenized language models and pronunciation lexicons. The resulting ASR models are comprehensively evaluated to answer the research questions regarding the impact of subword tokenization algorithms on language modeling complexity and on ASR performance. Our study highlights the strong performance of the hybrid S-BPE tokens, achieving a notable 10.6% word error rate (WER), which represents a substantial 16.8% improvement over the baseline word-level ASR system. The ablation study has revealed that the performance of S-BPE segmentation, which initially underperformed compared to syllable tokens with lower amounts of textual data for language modeling, exhibited steady improvement with the increase in LM training data. The extensive ablation study indicates that there is a limited advantage in raising the n-gram order of the lan

关键词： Subword tokens language modeling Open vocabulary Speech recognition Morphological complexity Malayalam language

来源：评论

学校读者我要写书评

暂无评论

LMTextSpotter: Towards Better Scene Text Spotting with language modeling in Transformer 18th

LMTextSpotter: Towards Better Scene Text Spotting with Langu...

引用

18th International Conference on Document Analysis and Recognition (ICDAR)

作者： Xia, Xin Ding, Guodong Li, Siyuan Zhejiang Univ Coll Comp Sci & Technol Hangzhou Peoples R China HiThink RoyalFlush Informat Network Co Ltd Hangzhou Peoples R China

ISBN: (纸本)9783031705489;9783031705496

End-to-end text spotting aims to build a unified framework for scene text detection and recognition. Recently, the DETR-like framework has attracted great attention. However, most of such studies merely focused on promoting the collaboration between two sub-tasks, without considering the importance of linguistic information in scene text. In this paper, we propose a novel end-to-end text spotting model, termed as LMTextSpotter, which introduces language modeling capability into the DETR-like text spotting framework. Specifically, we add an extra LM recognition branch besides the detection and commonly used CTC recognition branch in most existing frameworks, which is trained with the cross-entropy loss in autoregression manner. We backtrack the viterbi path given by CTC recognition branch to get an alignment between the queries and characters, so as to train LM recognition branch without relying on the character level annotations. After training phase, the LM recognition branch can predict the character classes independently of the CTC recognition branch. Besides, we exploit the position priors provided by rotated bounding boxes and adopt a task-specific query strategy for further improving the performance. Comparison results demonstrate the superiority of our LMTextSpotter over previous state-of-the-art methods. Ablation studies also prove the effectiveness of each component.

关键词： Scene Text Spotting Deformable Attention CTC-based Character Recognition language modeling

来源：评论

学校读者我要写书评

暂无评论

Improvements in language modeling, Voice Activity Detection, and Lexicon in OpenASR21 Low Resource languages 25th

Improvements in Language Modeling, Voice Activity Detection,...

引用

25th International Conference on Speech and Computer (SPECOM)

作者： Gupta, Vishwa Boulianne, Gilles Ctr Rech Informat Montreal CRIM Montreal PQ Canada

ISBN: (纸本)9783031483110;9783031483127

OpenASR21 evaluation was on 15 low resource languages and 3 case sensitive languages. During the evaluation, participants got significant reduction in word error rates (WER) with text downloaded from the internet for only the case sensitive languages, since the development and evaluation audio contained broadcast news. For the 15 low resource languages, participants showed only small gains for some of the languages. The reason is that the development and test set contain dialog between two people, which is very different from the primarily news texts and web pages available over the internet. Here, we show that training text translated from other OpenASR21 languages reduces the WER for many languages. During the evaluation, one team added words to the lexicon using a 3-gram phone language model, but they do not show what WER reduction they achieve. We show that adding new words in the lexicon from public text is beneficial for languages where the out-of-vocabulary rate is high, and outline conditions for reducing the WER. Adding an attention layer to the TDNN (time delay neural net) based voice activity detector reduced the WER for 17 out of the 18 languages. With all the improvements combined, we are getting lower word error rate for the development set for three languages (Farsi, Kazakh and Tamil) than the site that achieved the best error rate for all the languages during the evaluation period.

关键词： OpenASR21 Low-resource Speech recognition language modeling

来源：评论

学校读者我要写书评

暂无评论

A Statistical language modeling Framework for Extractive Summarization of Text Documents

引用

SN Computer Science 2023年第6期4卷 750页

作者： Gupta, Pooja Nigam, Swati Singh, Rajiv Department of Computer Science & Centre for Artificial Intelligence Banasthali Vidyapith Rajasthan India

The availability of a large collection of text documents on a variety of topics, such as tweets, web pages, news articles, and stories, in different languages. Due to these electronic documents, users get exhausted reading the entire document and face many difficulties finding relevant information. Text summarization is the solution for retrieving important information from multiple documents. The summarization process should be robust, language-independent, and able to generate summaries from multi-language corpora. With this aim, in this work, we propose a language-independent framework of extractive text summarization for the text documents, which are primarily based on English, and then the translated corpus of BBC, CNN, and DUC 2004 News stories from English to Hindi is exploited for this study. In the absence of a Hindi corpus for these datasets, we have used Google, Bing, and Systran machine translators for Hindi text generation to show the effectiveness of the proposed approach for a low resource language. The proposed method is based on the N-gram language model and statistical modeling using maximum likelihood estimation (MLE). Sentence ranking has been used to generate summaries of English and Hindi documents. We evaluate our model on the BBC, CNN, and DUC 2004 news stories datasets in both English and Hindi using the ROUGE metric at different levels to measure the performance of the generated summaries. Further, we compare the proposed method with other existing baseline and state-of-the-art methods, such as LexRank, TextRank, Lead, Luhn, LSA, and SumBasic. Experimental results of our approach indicate its effectiveness over existing summarization methods. © 2023, The Author(s), under exclusive licence to Springer Nature Singapore Pte Ltd.

关键词： Extractive summarization language modeling Machine translation Natural language processing ROUGE

来源：评论

学校读者我要写书评

暂无评论

Memory Augmented Lookup Dictionary based language modeling for Automatic Speech Recognition 24

Memory Augmented Lookup Dictionary based Language Modeling f...

引用

Interspeech Conference

作者： Feng, Yukun Tu, Ming Xia, Rui Huang, Chuanzeng Wang, Yuxuan ByteDance Inc San Jose CA 95110 USA

Recent studies have shown that using an external language Model (LM) benefits the end-to-end Automatic Speech Recognition (ASR). However, predicting tokens that appear less frequently in the training set is still quite challenging. The longtail prediction problems have been widely studied in many applications, but only been addressed by a few studies for ASR and LMs. In this paper, we propose a new memory augmented lookup dictionary based Transformer architecture for LM. The newly introduced lookup dictionary incorporates rich contextual information in training set, which is vital to correctly predict long-tail tokens. With intensive experiments on Chinese and English data sets, our proposed method is proved to outperform the baseline Transformer LM by a great margin on both word/character error rate and tail tokens error rate. This is achieved without impact on the decoding efficiency. Overall, we demonstrate the effectiveness of our proposed method in boosting the ASR decoding performance, especially for longtail tokens.

关键词： Automatic speech recognition language modeling rare words recognition long-tail recognition

来源：评论

学校读者我要写书评

暂无评论

Transformer-Based Deep Neural language modeling for Construct-Specific Automatic Item Generation

引用

PSYCHOMETRIKA 2022年第2期87卷 749-772页

作者： Hommel, Bjoern E. Wollang, Franz-Josef M. Kotova, Veronika Zacher, Hannes Schmukle, Stefan C. Univ Leipzig Leipzig Germany Magnolia Psychometr GMBH Leipzig Germany Tech Univ Munich Munich Germany

Algorithmic automatic item generation can be used to obtain large quantities of cognitive items in the domains of knowledge and aptitude testing. However, conventional item models used by template-based automatic item generation techniques are not ideal for the creation of items for non-cognitive constructs. Progress in this area has been made recently by employing long short-term memory recurrent neural networks to produce word sequences that syntactically resemble items typically found in personality questionnaires. To date, such items have been produced unconditionally, without the possibility of selectively targeting personality domains. In this article, we offer a brief synopsis on past developments in natural language processing and explain why the automatic generation of construct-specific items has become attainable only due to recent technological progress. We propose that pre-trained causal transformer models can be fine-tuned to achieve this task using implicit parameterization in conjunction with conditional generation. We demonstrate this method in a tutorial-like fashion and finally compare aspects of validity in human- and machine-authored items using empirical data. Our study finds that approximately two-thirds of the automatically generated items show good psychometric properties (factor loadings above .40) and that one-third even have properties equivalent to established and highly curated human-authored items. Our work thus demonstrates the practical use of deep neural networks for non-cognitive automatic item generation.

关键词： automatic item generation natural language processing deep learning neural networks language modeling

来源：评论

学校读者我要写书评

暂无评论

A Novel Approach for Text Generation using RNN for language modeling 3

A Novel Approach for Text Generation using RNN for Language ...

引用

3rd International Conference on Innovative Mechanisms for Industry Applications, ICIMIA 2023

作者： Kumar, P. Manikandan, S. Kishore, R. Rajalakshmi Engineering College Department of CSE Chennai India

ISBN: (纸本)9798350343632

In recent years, many research works are initiated in the domain of text generation as it is the most important and needed task in natural language processing. The text generation process has various applications including chatbots, social media, language translation, news, movie script writing, and poetry comprehension. This research study provides a detailed study on text generation with the help of SRNN using LSTM and GRU for language modeling and also to generate coherent and contextually relevant text equivalent to human creativity level. The techniques such as teacher forcing, beam search, and temperature scaling are used to explore the good quality and diversity of the generated text. This study utilizes large-scale text including news articles, books and blogs to train and evaluate RNN based language models. Evaluation metrics such as perplexity, BLEU score, human evaluation to assess the fluency, coherence and overall quality of the generated text. This research study investigates the performance of RNN-based language models based on hyperparameters and the impact on different architectural variations and training strategies. This study presents several automated text generation techniques that shorten the time required to manually type and produce lengthy texts. © 2023 IEEE.

关键词： Gated Recurrent Unit language modeling Long Short Term Memory Recurrent Neural Network Text Generation

来源：评论

学校读者我要写书评

暂无评论

Test Sample Selection for Handwriting Recognition Through language modeling 10th

Test Sample Selection for Handwriting Recognition Through La...

引用

10th Iberian Conference on Pattern Recognition and Image Analysis (IbPRIA)

作者： Rosello, Adrian Ayllon, Eric Valero-Mas, Jose J. Calvo-Zaragoza, Jorge Univ Alicante Dept Software & Comp Syst Alicante Spain

ISBN: (纸本)9783031048814;9783031048807

When it comes the need to automatically recognize hand- written information, there may be a scenario in which the capture device (such as the camera of a mobile phone) retrieves several samples-a burst-of the same content. Unlike general classification scenarios, combining different inputs is not straightforward for state-of-the-art handwriting recognition approaches based on neural end-to-end formulations. Moreover, since not all images within the burst may depict enough quality for contributing in the overall recognition task, it is hence necessary to select the appropriate subset. In this work, we propose a pilot study which addresses this scenario in which there exists a burst of pictures of the same manuscript to transcribe. Within this context, we present a straightforward strategy to select a single candidate out of the elements of the burst which achieves the best possible transcription. Our hypothesis is that the best source image is that whose transcription most likely matches a language Model (LM) estimated on the application domain-in this work implemented by means of n-grams. Our strategy, therefore, processes all the images of the burst and selects the transcription with the highest a priori probability according to the LM. Our experiments recreate this scenario in two typical benchmarks of Handwritten Text Recognition (HTR) and Handwritten Music Recognition (HMR), validating the goodness of our proposal and leaving room for promising future work.

关键词： Handwriting recognition language modeling Convolutional recurrent neural network

来源：评论

学校读者我要写书评

暂无评论

A BERT-based language modeling Framework 23

A BERT-based Language Modeling Framework

引用

Interspeech Conference

作者： Chien, Chin-Yueh Chen, Kuan-Yu KenKone Med Co Ltd Taipei Taiwan Natl Taiwan Univ Sci & Technol Taipei Taiwan

Deep learning has brought considerable changes and created a new paradigm in many research areas, including computer vision, speech processing, and natural language processing. In the context of language modeling, recurrent-based language models and word embedding methods have been pivotal studies in the past decade. Recently, pre-trained language models have attracted widespread attention due to their enormous success in many challenging tasks. However, only a dearth of works concentrates on creating novel language models based on the pre-trained models. In order to bridge the research gap, we take the bidirectional encoder representations from Transformers (BERT) model as an example to explore novel uses of a pre-trained model for language modeling. More formally, this paper proposes a set of BERT-based language models, and a neural-based dynamic adaptation method is also introduced to combine these language models systematically and methodically. We conduct comprehensive studies on three datasets for the perplexity evaluation. Experiments show that the proposed framework achieves 11%, 39%, and 5% relative improvements over the baseline model for Penn Treebank, Wikitext 2, and Tedlium Release 2 corpora, respectively. Besides, when applied to rerank n-best lists from a speech recognizer, our framework also yields promising results compared with baseline systems.

关键词： pre-trained language model BERT language modeling dynamic adaptation

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：