首页 / 专利库 / 人工智能 / 机器学习 / 监督学习 / 相似性学习 / Recognizing system for handwritten numeral

Recognizing system for handwritten numeral

阅读:694发布:2021-12-07

专利汇可以提供Recognizing system for handwritten numeral专利检索,专利查询,专利分析的服务。并且PURPOSE: To improve the recognition rate by deriving an output value to learning data of a neural network which is allowed to learn primary inter-pattern similarity calculated, based on an inter-pattern distance, and allowing it to learn secondary inter-pattern similarity calculated, based on its output value.
CONSTITUTION: Output values 34, 35 to learning data of a neural network which is allowed to learn primary inter-pattern similarity calculated, based on an inter-pattern distance of a feature pattern of an input numeral and a standard pattern are derived. Subsequently, it is allowed to learn secondary inter-pattern similarity calculated, based on the output values 34, 35, and neural networks 11, 12 are allowed to learn more faithfully the inter-pattern similarity by which a correct solution cannot always be obtained by only calculating the inter- pattern distance. In such a way, the recognition is improved.
COPYRIGHT: (C)1992,JPO&Japio,下面是Recognizing system for handwritten numeral专利的具体信息内容。

高效检索全球专利

专利汇是专利免费检索,专利查询,专利分析-国家发明专利查询检索分析平台,是提供专利分析,专利查询,专利检索等数据服务功能的知识产权数据服务商。

我们的产品包含105个国家的1.26亿组数据,免费查、免费专利分析。

申请试用

分析报告

专利汇分析报告产品可以对行业情报数据进行梳理分析,涉及维度包括行业专利基本状况分析、地域分析、技术分析、发明人分析、申请人分析、专利权人分析、失效分析、核心专利分析、法律分析、研发重点分析、企业专利处境分析、技术处境分析、专利寿命分析、企业定位分析、引证分析等超过60个分析角度,系统通过AI智能系统对图表进行解读,只需1分钟,一键生成行业专利分析报告。

申请试用

QQ群二维码
意见反馈