Scholarlink

<p>Taking into account that the results obtained with a BiLSTM network in the recognition of handwritten mathematical symbols can improve depending on the number of features used in the training and the size of the training set, in this article, a model that uses a set of features online which includes a new feature based on projections is proposed and an alternative that allows increasing the size of the training set for the BiLSTM network is presented obtaining competitive results. Additionally, hybrid models that combine CNN and BiLSTM networks are presented, with good results.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Classification of handwritten mathematical symbols with a BiLSTM network using a new online feature and permuting strokes

  • 期刊论文

  • 发表日期：

    2025年9月12日

查看全文
  • Mauricio Maca,
  • Rosana Pérez

摘要

Taking into account that the results obtained with a BiLSTM network in the recognition of handwritten mathematical symbols can improve depending on the number of features used in the training and the size of the training set, in this article, a model that uses a set of features online which includes a new feature based on projections is proposed and an alternative that allows increasing the size of the training set for the BiLSTM network is presented obtaining competitive results. Additionally, hybrid models that combine CNN and BiLSTM networks are presented, with good results.