Bank of America Corporation
Reinforcement learning based document scanner
Last updated:
Abstract:
Apparatus and methods for transformation of a digital scanner image using machine-learning algorithms are provided. The apparatus and methods may include a portable USB device configured for connection to a scanner port. The device may access and store a scanned digital image captured by the scanner. A device processor may use OCR to generate an editable PDF file and use one or more machine-learning algorithms to apply auto-corrections to the PDF file. The processor may communicate with a user interface configured to display each line from the scanned digital image in line with the corresponding auto-corrected text. The user interface may receive separate inputs accepting each line of auto-corrected text. Auto-correction acceptance data may be transmitted to the device processor. Each accepted auto-correction may be associated with a quantified value. A machine-learning algorithm may be configured to maximize a total value for auto-corrections in a scanned document.
Utility
7 Nov 2017
31 Mar 2020