TONE MODELING BASED ON HIDDEN CONDITIONAL RANDOM FIELDS AND DISCRIMINATIVE MODEL WEIGHT TRAINING

The use of hidden conditional random fields （HCRFs） for tone modeling is explored. The tone recognition performance is improved using HCRFs by taking advantage of intra-syllable dynamic, inter-syllable dynamic and duration features. When the tone model is integrated into continuous speech recognition, the discriminative model weight training （DMWT） is proposed. Acoustic and tone scores are scaled by model weights discriminatively trained by the minimum phone error （MPE） criterion. Two schemes of weight training are evaluated and a smoothing technique is used to make training robust to overtraining problem. Experiments show that the accuracies of tone recognition and large vocabulary continuous speech recognition （LVCSR） can be improved by the HCRFs based tone model. Compared with the global weight scheme, continuous speech recognition can be improved by the discriminative trained weight combinations.

Conditional Random Field Tracking Model Based on a Visual Long Short Term Memory Network

In dense pedestrian tracking,frequent object occlusions and close distances between objects cause difficulty when accurately estimating object trajectories.In this study,a conditional random field tracking model is established by using a visual long short term memory network in the three-dimensional(3D)space and the motion estimations jointly performed on object trajectory segments.Object visual field information is added to the long short term memory network to improve the accuracy of the motion related object pair selection and motion estimation.To address the uncertainty of the length and interval of trajectory segments,a multimode long short term memory network is proposed for the object motion estimation.The tracking performance is evaluated using the PETS2009 dataset.The experimental results show that the proposed method achieves better performance than the tracking methods based on the independent motion estimation.

Exploiting PLSA model and conditional random field for refining image annotation

This paper presents a new method for refining image annotation by integrating probabilistic la- tent semantic analysis （PLSA） with conditional random field （CRF）. First a PLSA model with asymmetric modalities is constructed to predict a candidate set of annotations with confidence scores, and then model semantic relationship among the candidate annotations by leveraging conditional ran- dom field. In CRF, the confidence scores generated lay the PLSA model and the Fliekr distance be- tween pairwise candidate annotations are considered as local evidences and contextual potentials re- spectively. The novelty of our method mainly lies in two aspects ： exploiting PLSA to predict a candi- date set of annotations with confidence scores as well as CRF to further explore the semantic context among candidate annotations for precise image annotation. To demonstrate the effectiveness of the method proposed in this paper, an experiment is conducted on the standard Corel dataset and its re- sults are ＇compared favorably with several state-of-the-art approaches.

An Image Segmentation Algorithm Based on a Local Region Conditional Random Field Model

To reduce the computation cost of a combined probabilistic graphical model and a deep neural network in semantic segmentation, the local region condition random field (LRCRF) model is investigated which selectively applies the condition random field (CRF) to the most active region in the image. The full convolutional network structure is optimized with the ResNet-18 structure and dilated convolution to expand the receptive field. The tracking networks are also improved based on SiameseFC by considering the frame relations in consecutive-frame traffic scene maps. Moreover, the segmentation results of the greyscale input data sets are more stable and effective than using the RGB images for deep neural network feature extraction. The experimental results show that the proposed method takes advantage of the image features directly and achieves good real-time performance and high segmentation accuracy.

Rockhead profile simulation using an improved generation method of conditional random field

Rockhead profile is an important part of geological profiles and can have significant impacts on some geotechnical engineering practice,and thus,it is necessary to establish a useful method to reverse the rockhead profile using site investigation results.As a general method to reflect the spatial distribution of geo-material properties based on field measurements,the conditional random field(CRF)was improved in this paper to simulate rockhead profiles.Besides,in geotechnical engineering practice,measurements are generally limited due to the limitations of budget and time so that the estimation of the mean value can have uncertainty to some extent.As the Bayesian theory can effectively combine the measurements and prior information to deal with uncertainty,CRF was implemented with the aid of the Bayesian framework in this study.More importantly,this simulation procedure is achieved as an analytical solution to avoid the time-consuming sampling work.The results show that the proposed method can provide a reasonable estimation about the rockhead depth at various locations against measurement data and as a result,the subjectivity in determining prior mean can be minimized.Finally,both the measurement data and selection of hyper-parameters in the proposed method can affect the simulated rockhead profiles,while the influence of the latter is less significant than that of the former.

Power entity recognition based on bidirectional long short-term memory and conditional random fields

With the application of artificial intelligence technology in the power industry,the knowledge graph is expected to play a key role in power grid dispatch processes,intelligent maintenance,and customer service response provision.Knowledge graphs are usually constructed based on entity recognition.Specifically,based on the mining of entity attributes and relationships,domain knowledge graphs can be constructed through knowledge fusion.In this work,the entities and characteristics of power entity recognition are analyzed,the mechanism of entity recognition is clarified,and entity recognition techniques are analyzed in the context of the power domain.Power entity recognition based on the conditional random fields (CRF) and bidirectional long short-term memory (BLSTM) models is investigated,and the two methods are comparatively analyzed.The results indicated that the CRF model,with an accuracy of 83%,can better identify the power entities compared to the BLSTM.The CRF approach can thus be applied to the entity extraction for knowledge graph construction in the power field.

A CONDITIONAL RANDOM FIELDS APPROACH TO BIOMEDICAL NAMED ENTITY RECOGNITION

Named entity recognition is a fundamental task in biomedical data mining. In this letter, a named entity recognition system based on CRFs （Conditional Random Fields） for biomedical texts is presented. The system makes extensive use of a diverse set of features, including local features, full text features and external resource features. All features incorporated in this system are described in detail, and the impacts of different feature sets on the performance of the system are evaluated. In order to improve the performance of system, post-processing modules are exploited to deal with the abbreviation phenomena, cascaded named entity and boundary errors identification. Evaluation on this system proved that the feature selection has important impact on the system performance, and the post-processing explored has an important contribution on system performance to achieve better resuits.

Standardization of Robot Instruction Elements Based on Conditional Random Fields and Word Embeddin

Natural language processing has got great progress recently. Controlling robots with spoken natural language has become expectable. With the reliability problem of this kind of control in mind a confirmation process of natural language instruction should be included before carried out by the robot autonomously and the prototype dialog system was designed thus the standardization problem was raised for the natural and understandable language interaction. In the application background of remotely navigating a mobile robot inside a building with Chinese natural spoken language considering that as an important navigation element in instructions a place name can be expressed with different lexical terms in spoken language this paper proposes a model for substituting different alternatives of a place name with a standard one (called standardization). First a CRF (Conditional Random Fields) model is trained to label the term required be standardized then a trained word embedding model is to represent lexical terms as digital vectors. In the vector space similarity of lexical terms is defined and used to find out the most similar one to the term picked out to be standardized. Experiments show that the method proposed works well and the dialog system responses to confirm the instructions are natural and understandable.

基于BERT-BiLSTM-CRF模型的畜禽疫病文本分词研究

针对畜禽疫病文本语料匮乏、文本内包含大量疫病名称及短语等未登录词问题,提出了一种结合词典匹配的BERT-BiLSTM-CRF畜禽疫病文本分词模型。以羊疫病为研究对象,构建了常见疫病文本数据集,将其与通用语料PKU结合,利用BERT(Bidirectional encoder representation from transformers)预训练语言模型进行文本向量化表示;通过双向长短时记忆网络(Bidirectional long short-term memory network,BiLSTM)获取上下文语义特征;由条件随机场(Conditional random field,CRF)输出全局最优标签序列。基于此,在CRF层后加入畜禽疫病领域词典进行分词匹配修正,减少在分词过程中出现的疫病名称及短语等造成的歧义切分,进一步提高了分词准确率。实验结果表明,结合词典匹配的BERT-BiLSTM-CRF模型在羊常见疫病文本数据集上的F1值为96.38%,与jieba分词器、BiLSTM-Softmax模型、BiLSTM-CRF模型、未结合词典匹配的本文模型相比,分别提升11.01、10.62、8.3、0.72个百分点,验证了方法的有效性。与单一语料相比,通用语料PKU和羊常见疫病文本数据集结合的混合语料,能够同时对畜禽疫病专业术语及疫病文本中常用词进行准确切分,在通用语料及疫病文本数据集上F1值都达到95%以上,具有较好的模型泛化能力。该方法可用于畜禽疫病文本分词。

融合多尺度CNN和CRF的通用细粒度事件检测

事件检测是自然语言处理领域中事件抽取的主要任务之一,它旨在从众多非结构化信息中自动提取出结构化的关键信息.现有的方法存在特征提取不全面、特征分布不均等情况.为了提高事件检测的准确率,提出了一种融合BERT预训练模型与多尺度CNN的神经网络模型(BMCC,BERT+Multi-scale CNN+CRF).首先通过BERT(Bidirectional Encoder Representations from Transformers)预训练模型来进行词向量的嵌入,并利用其双向训练的Transformer机制来提取序列的状态特征;其次使用不同尺度的卷积核在多个卷积通道中进行卷积训练,以此来提取不同视野的语义信息,丰富其语义表征.最后将BIO机制融入到条件随机场(CRF)来对序列进行标注,实现事件的检测.实验结果表明,所提出的模型在MAVEN数据集上的F1值为65.17%,表现了该模型的良好性能.

基于BERT-BiLSTM-CRF模型的油气领域命名实体识别

针对油气领域知识图谱构建过程中命名实体识别使用传统方法存在实体特征信息提取不准确、识别效率低的问题,提出了一种基于BERT-BiLSTM-CRF模型的命名实体识别研究方法。该方法首先利用BERT(bidirectional encoder representations from transformers)预训练模型得到输入序列语义的词向量;然后将训练后的词向量输入双向长短期记忆网络(bi-directional long short-term memory,BiLSTM)模型进一步获取上下文特征;最后根据条件随机场(conditional random fields,CRF)的标注规则和序列解码能力输出最大概率序列标注结果,构建油气领域命名实体识别模型框架。将BERT-BiLSTM-CRF模型与其他2种命名实体识别模型(BiLSTM-CRF、BiLSTM-Attention-CRF)在包括3万多条文本语料数据、4类实体的自建数据集上进行了对比实验。实验结果表明,BERT-BiLSTM-CRF模型的准确率(P)、召回率(R)和F_(1)值分别达到91.3%、94.5%和92.9%,实体识别效果优于其他2种模型。

基于ERNIE-BiGRU-Attention-CRF的电子病历命名实体识别方法

目的/意义改善中文电子病历命名实体识别模型的性能,更好地开展医疗信息的组织和挖掘。方法/过程构建ERNIE-BiGRU-Attention-CRF中文电子病历命名实体识别模型,首先采用ERNIE1.0预训练模型生成具有语义特征的词向量,然后利用BiGRU捕获全局语义特征与语法结构特征,通过Attention机制进一步增强语义特征的捕获,最后连接CRF解码层输出全局概率最大的标签序列。结果/结论在公开的医疗文本数据集CCKS2017开展对比实验、消融实验,利用生成的模型进行实例分析,取得较好的识别效果。

基于改进Bi-LSTM-CRF的农业问答系统研究

针对农业领域问答系统面临的实体识别困难的问题,提出一种基于改进Bi-LSTM-CRF的实体识别方法。首先通过BERT预训练模型的预处理,生成基于上下文信息的词向量,然后将训练出的词向量输入Bi-LSTM-CRF做进一步的训练处理,最后,利用Python的Django框架设计农业领域的实体识别、实体查询、农知问答等子系统。经过试验对比,所提出的改进的Bi-LSTM-CRF在农业信息领域具有更好的实体识别能力,在农业信息语料库上的精确率、召回率和F1值分别为93.23%、91.08%和92.16%。实现农业领域实体识别和农业信息问答的知识图谱网站演示,对农业信息化的发展具有重要意义。

基于BERT-BILSTM-CRF模型的电力行业事故文本智能分析

为解决电力行业事故报告文本较长、语义复杂,难以进行有效文本识别问题,提出1种以BERT作为底层的预训练模型,并设计1种双重注意力机制编码器,结合BILSTM-CRF深度挖掘事故文本语义特征,从而实现文本智能分析。首先构建电力词典,通过对BERT预训练,进行BIO标注,然后引入BILSTM-CRF模型实现对文本标签智能分类,最后将该模型与现行其他4种深度学习模型进行对比。研究结果表明:该模型智能识别精确率、召回率及F 1值(查准率)均达到约97%,较其他4种模型中效果最好的模型分别提高0.02,0.03,0.02。研究结果可为电力行业事故报告文本分析提供1种新思路。

基于HMM+CRF词性标注的实体抽取方法

基于HMM+CRF词性标注的实体抽取方法从词性标注入手,对待处理文本先进行词性标注;然后根据文本的词性将实体抽取出来,在传统的CRF词性标注模型上增加一层HMM模型,提高实体抽取的精确度;最后在人民日报语料上进行实验,验证了准确率分别在基于HMM实体抽取模型和基于CRF实体抽取模型的基础上提高了2.1%和0.3%。

融合汉字部首的BERT-BiLSTM-CRF中医医案命名实体识别模型

目的/意义研究提取中医医案中医疗术语的方法,实现医案自动结构化,为医案知识发现提供结构化数据。方法/过程提出一种BERT结合长短期记忆人工神经网络、条件随机场和部首特征的深度学习命名实体识别模型,在BERT词向量中嵌入汉字部首,采用双向长短期记忆人工神经网络提取实体特征,使用条件随机场进行序列预测。将人工标注的400份共计5万余字的医案按照3∶1划分为训练集和测试集,使用该模型识别中医医案中的身体部位、药物、症状、疾病4类命名实体。结果/结论该模型在测试集F 1值为84.81%,优于其他未嵌入部首的模型,表明该模型能够更有效地识别中医医案中的命名实体,更好地结构化医案。

基于Wobert与对抗学习的中文命名实体识别

由于自然语言处理(NLP)将中文命名实体识别(NER)任务建模为序列标注任务,将文本中每个字符映射至一个标签,每个字符相对独立且信息有限,因此在NER领域词汇信息的加入能够解决字符间缺乏联系的问题。针对现有中文NER模型多需要额外构建词汇表、提取词汇信息方式繁琐、词级嵌入与字级嵌入因来源不同导致信息难以融合的问题,提出一种基于Wobert与对抗学习的中文NER模型ALWAE-BiLSTM-CRF。与传统预训练模型相比,Wobert预训练模型在预训练阶段就已将文本分词,充分学习了词与字两个层次的信息,因此ALWAE-BiLSTM-CRF通过Wobert预训练模型获取字符词向量,再使用Wobert分词器获取预训练模型中已存在的词汇向量,接着使用BiLSTM模型获取两者的时序信息,随后通过多头注意力机制将词汇级别的信息要素融入字符词向量,同时通过对抗学习攻击生成对抗样本以增强模型泛化性,最后使用条件随机场(CRF)层对结果进行约束,获得最佳的预测序列。在Resume数据集与瓷器领域的自建数据集Porcelain上进行对比实验与消融实验,结果表明,ALWAE-BiLSTM-CRF模型的F1值分别达到97.21%与85.7%,证明了其在中文NER任务中的有效性。

基于命名实体识别的水电工程施工安全规范实体识别模型

为准确识别水电工程施工安全规范实体,通过预训练模型中双向编码器表征法(BERT)挖掘文本中丰富的语义信息,利用双向长短期记忆神经网络(BILSTM)提取规范实体语义特征,依靠条件随机场(CRF)分析实体之间的依赖关系,构建水电工程施工安全规范的命名实体识别模型;以《水利水电工程施工安全防护技术规范》(SL714—2015)为例,计算命名实体识别模型精确率。结果表明:BERT-BILSTM-CRF模型准确率为94.35%,相比于3种传统方法,准确率显著提高。研究成果有助于水电工程施工安全规范知识智能管理,为施工安全隐患智能判别提供支撑。

基于BERT-BiLSTM-CRF的电力集控安全隐患数据处理

为了提高电力集控系统安全隐患数据处理的效果,提出一种基于来自变换器的双向编码器表示-双向长短期记忆网络-条件随机场(Bidirectional Encoder Representations from Transformers-Bidirectional Long Short Term Memory-Conditional Random Fields,BERT-BiLSTM-CRF)的电力集控安全隐患数据处理方法。构建电力集控隐患数据检测模型,应用改进长短时记忆网络(Long Short Term Memory,LSTM)来构建电力集控安全隐患数据修复网络,实现电力集控安全隐患数据处理。实验结果表明,采用所提方法能够更好地完成电力集控安全隐患数据检测与修复,应用效果较好。

基于改进DeeplabV3+的水面多类型漂浮物分割方法研究

【目的】为解决传统图像处理方法鲁棒性差、常用深度学习检测方法无法准确识别大片漂浮物的边界等问题,【方法】提出一种基于改进DeeplabV3+的水面多类型漂浮物识别的语义分割方法,提高水面漂浮的识别能力。对所收集实际水面漂浮物进行分类,采用自制数据集进行对比试验。算法选择xception网络作为主干网络以获得初步漂浮物特征,在加强特征提取网络部分引入注意力机制以强调有效特征信息,在后处理阶段加入全连接条件随机场模型,将单个像素点的局部信息与全局语义信息融合。【结果】对比图像分割性能指标,改进后的算法mPA(Mean Pixel Accuracy)提升了5.73%,mIOU(Mean Intersection Over Union)提升了4.37%。【结论】相比于其他算法模型,改进后的DeeplabV3+算法对漂浮物特征的获取能力更强,同时能获得丰富的细节信息以更精准地识别多类型水面漂浮物的边界与较难分类的漂浮物,在对多个水库场景测试后满足实际水域环境中漂浮物检测的需求。