Unstructured data extraction system using multi head attention and a novel language model /

IPI.jpg

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.

Creator

V B, Kirubanand.

Publisher

Intellectual Property India

Date

2021

Language

English

Type

Patent

Collection

Citation

V B, Kirubanand., “Unstructured data extraction system using multi head attention and a novel language model /,” CHRIST (Deemed To Be University) Institutional Repository, accessed December 8, 2023, http://archives.christuniversity.in/items/show/2549.

Output Formats