首页 > AI前沿 > Domain Adaptation Based Pipeline for Character Classification and Handwritten Text Recognition

Domain Adaptation Based Pipeline for Character Classification and Handwritten Text Recognition

arXiv机器学习 2026-11-13 17:40 5 阅读 查看原文
用户现在需要把给定的英文新闻正文排版成语义化HTML片段,首先得按照规则来。首先看正文是一段一段的?不对,先理清楚结构:首先开头的段落是讲问题和提出的pipeline,然后有三步,对吧?首先第一个段落应该是开头到“distribution gap between source and target domains.”?不对,等下原文的结构:首先第一句到提出pipeline,然后说three main steps,接下来三个步骤,然后是效果、评估、结果、代码。 哦不对,首先,每个自然段落用p包裹,然后有没有小节?比如可以设h2?不对,原文是连贯的?等下看:首先第一部分是背景和提出的方法概述:

Character classification and handwritten text recognition...between source and target domains.

然后接下来是三个步骤的部分,要不要用h3?不对,或者,哦,等下,原文里说Our pipeline follows three main steps. 然后三个步骤,那可以?不对,首先不要增删内容,所以结构要对应。 等下逐句来: 第一个自然段:Character classification and handwritten text recognition, particularly in specialized tasks, often suffer from limited labeled data, while a large amount of unlabeled data remains unused. To address this challenge, we propose a novel unsupervised domain adaptation-based pipeline to reduce the distribution gap between source and target domains. 这部分是一个段落,用

包裹。 然后接下来,是不是可以有个h2?比如

Pipeline Steps

?不对啊,不能增内容啊!哦对哦,规则说保留原文所有信息,不得增删改内容,所以不能自己加标题文字!哦对,我刚才差点错了。那怎么弄?哦不对,原文里有没有可以划分小节的?等下原文的结构: 首先背景和方法提出:第一段。然后接下来讲三个步骤的内容,然后是方法的效果、评估、结果、代码。不对,那有没有可以用h3的?哦不对,不能自己加文字,那是不是直接按自然段?等下看原文的自然段:用户给的正文是一整块?不对,看用户给的正文: “Character classification and handwritten text recognition, particularly in specialized tasks, often suffer from limited labeled data, while a large amount of unlabeled data remains unused. To address this challenge, we propose a novel unsupervised domain adaptation-based pipeline to reduce the distribution gap between source and target domains. Our pipeline follows three main steps. In the first step, we pretrained an expert model on source data, to learn a strong general representation. In the second step, we perform unsupervised domain adaptation to leverage unlabeled target data, adapting the model to the specific characteristics of the target domain. In the third step, we trained the expert model on source data in the target style. By aligning the feature distributions between the source and target, our method enables the model to generalize effectively across diverse datasets. The unsupervised training increases the computation time only during training not in inference. We evaluate the pipeline on both character-level and historical handwritten line recognition tasks, demonstrating its flexibility. We establish a new state-of-the-art on the Safran-MNIST-DLS dataset defined in Track 2 of the DAGECC competition with 76.17% Macro F1-score and achieve competitive results on each of the specific datasets of READ 2018 without using the labels of the target data. Code is available at https://github.com/simon-corbi/DRANet-for-classification-and-htr” 哦,原来用户给的是一段?不对,但是规则说长文用h2/h3划分小节,但是不能增删内容,那怎么弄?哦,不对,是不是可以把结构上的部分划分,但是标题内容是原文里的?不对,原文里没有标题啊。哦不对,等下,有没有可能我理解错了,就是如果原文有层级的内容就用,没有的话就按自然段?不对,那这里的三个步骤,是不是可以用列表?哦对!规则说列表内容用ul/li。哦对!原文里说“three main steps”,后面是三个步骤,所以那三个步骤可以做成列表对吧? 等下我理一下: 首先第一个

:内容是背景和提出pipeline,到“Our pipeline follows three main steps.”?不对,不对,“Our pipeline follows three main steps.”是引出步骤的,然后三个步骤: 第一个步骤:In the first step, we pretrained an expert model on source data, to learn a strong general representation. 第二个:In the second step, we perform unsupervised domain adaptation to leverage unlabeled target data, adapting the model to the specific characteristics of the target domain. 第三个:In the third step, we trained the expert model on source data in the target style. 那这三个是不是可以做成

  • ?对,因为是步骤列表,属于列表内容对吧。 然后接下来,

    的内容是:By aligning the feature distributions between the source and target, our method enables the model to generalize effectively across diverse datasets. The unsupervised training increases the computation time only during training not in inference. We evaluate the pipeline on both character-level and historical handwritten line recognition tasks, demonstrating its flexibility. We establish a new state-of-the-art on the Safran-MNIST-DLS dataset defined in Track 2 of the DAGECC competition with 76.17% Macro F1-score and achieve competitive results on each of the specific datasets of READ 2018 without using the labels of the target data. 然后代码的那句,是不是可以单独?不对,或者把代码那句里的链接?不对,规则说不要增删改,所以保留。哦,还有,重要句子可以用strong,比如提出方法的那句?或者新SOTA的那句?可以加strong对吧,只要是原文的内容,只是加标签。 等下,我再梳理结构,不要加任何原文没有的内容,所以标题?哦不对,原文里没有小节标题,那是不是不用h2/h3?不对,规则说“长文用

    /

    划分小节”,但是如果原文没有小节标题的话,是不是不能硬加?哦不对,等下,会不会我错了,是不是可以把有层级的部分用h3,但是内容是原文里的?比如,有没有?哦,不对,原文是连贯的,没有小标题,那可能不需要用h2/h3?不对,等下再看规则,规则3是列表、引用、代码、重要句子的标签,规则2是每个自然段p,长文用h2/h3划分,那如果原文没有小节的话,就不用?不对,等下,我再看原文的结构: 首先,第一部分:问题和提出的pipeline概述:

    Character classification and handwritten text recognition, particularly in specialized tasks, often suffer from limited labeled data, while a large amount of unlabeled data remains unused. To address this challenge, we propose a novel unsupervised domain adaptation-based pipeline to reduce the distribution gap between source and target domains. Our pipeline follows three main steps.

    然后是三个步骤,列表:
    • In the first step, we pretrained an expert model on source data, to learn a strong general representation.
    • In the second step, we perform unsupervised domain adaptation to leverage unlabeled target data, adapting the model to the specific characteristics of the target domain.
    • In the third step, we trained the expert model on source data in the target style.
    然后下一个

    ,讲方法的效果和评估:

    By aligning the feature distributions between the source and target, our method enables the model to generalize effectively across diverse datasets. The unsupervised training increases the computation time only during training not in inference. We evaluate the pipeline on both character-level and historical handwritten line recognition tasks, demonstrating its flexibility. We establish a new state-of-the-art on the Safran-MNIST-DLS dataset defined in Track 2 of the DAGECC competition with 76.17% Macro F1-score and achieve competitive results on each of the specific datasets of READ 2018 without using the labels of the target data.

    然后最后一个

    是代码的那句:

    Code is available at https://github.com/simon-corbi/DRANet-for-classification-and-htr

    ?不对,规则里代码用
    ?哦不对,那个是URL,算不算代码?哦,规则里的代码是代码片段?还是URL也可以用code?不对,等下原文里的URL是代码地址,那是不是可以不用pre?因为pre是保留格式的,URL是行内的话,用code就行?不对,等下看规则3:“代码用
    ”,那如果是行内的链接,是不是不需要pre?哦,对,pre是块级的,行内的代码用code就行。或者,有没有问题?不对,再看,原文的代码地址那句,就是“Code is available at https://github.com/simon-corbi/DRANet-for-classification-and-htr”,那是不是可以写成

    Code is available at https://github.com/simon-corbi/DRA