Unstructured data extraction system using multi head attention and a novel language model /
Title
Unstructured data extraction system using multi head attention and a novel language model /
Subject
A system 100 for Offline handwritten text recognition (HTR) of a scanned handwritten text input image
Description
Patent Number: 202141056398, Applicant: K. P. Kavitha.
A system 100 for Offline handwritten text recognition (HTR) of a scanned handwritten text input image leveraging Modern Deep Recurrent Neural Network (RNN). System 100 comprises (RNN) is proposed with the help of the present's embodiments disclosure (RNN). A cursive eliminated handwritten text image is mapped to a multi-head attention-based sequence-to-sequence learning applying the beam search technique and employing an RNN-based variable-length encoder-decoder architecture.
A system 100 for Offline handwritten text recognition (HTR) of a scanned handwritten text input image leveraging Modern Deep Recurrent Neural Network (RNN). System 100 comprises (RNN) is proposed with the help of the present's embodiments disclosure (RNN). A cursive eliminated handwritten text image is mapped to a multi-head attention-based sequence-to-sequence learning applying the beam search technique and employing an RNN-based variable-length encoder-decoder architecture.
Creator
Paulose, Joy.
Publisher
Intellectual Property India
Date
2021
Language
English
Type
Patent
Collection
Citation
Paulose, Joy., “Unstructured data extraction system using multi head attention and a novel language model /,” CHRIST (Deemed To Be University) Institutional Repository, accessed December 22, 2024, https://archives.christuniversity.in/items/show/2550.