Improving our technology to increase automation

Improving our technology to increase automation

In Serimag we consider as important the development of the different projects through the application of the different Artificial Intelligence and Neural Networks technologies that we know, as the investment in R&D to continue evolving these techniques.

TwitterLinkedIn

In Serimag we consider as important the development of the different projects through the application of the different Artificial Intelligence and Neural Networks technologies that we know, as the investment in R&D to continue evolving these techniques. In each project evaluation and in each production process Serimag incorporates the most appropriate technology to the needs of the service and our customer thanks to the professionalism and knowledge of all technicians and project managers. We live in a rapidly changing technological world where new techniques are continually being developed to improve existing processes. At Serimag we want to be part of this technological revolution.

COLLABORATION WITH BARCELONA SUPERCOMPUTING CENTER (BSC)

After months of joint research between Serimag and the Barcelona Supercomputing Center (BSC-UPC), the Barcelona Supercomputing Center (BSC-UPC) a new unique graphical-textual classification system based on neural networks has been developed, which Serimag will promote in all our document classification processes.

Traditional document classification systems based on rigid, predefined templates and rules have long since been superseded in highly automated production environments such as the ones in which we operate. TAAD offers. However, the standard in document intelligence-based automation still processes a document from a graphical and textual point of view independently, merging both verdicts into a final classification system:

These classification systems, despite being based on Machine Learning, require intermediate and explicit layers between one processing and another, which inevitably prevent the extraction of the maximum benefit from the intelligent recognition engines and require manual parameterization with the consequent maintenance.

From Serimag we started a collaboration project with BSC a year ago, based on the following hypothesis: there must be an architecture of neural network capable of performing complete classification (both graphical and textual) in an integrated manner and without the need for parametric coupling modules, understanding the document as a whole.

Today we can affirm that the hypothesis was true and that it does indeed yield substantial improvements in terms of time and production results.

APPLICATION IN PROJECTS WITH TESSI-GRADED

At Serimag we provide a constantly evolving service. During the life of a process we periodically evaluate potential technological improvements resulting from our R&D laboratory that will improve automation rates for mutual benefit to both the customer and Serimag. With three years of joint collaboration for the automation of document processes, Serimag has begun the introduction of this new technology in the production processes that we share with Tessi-Graddo .

Thus, this new neural network model has been implemented in current production models together with Tessi-Graddo, which have managed to reduce by half the rejection of previous models, increasing the accuracy to over 99%:

We are already working on new systems also applicable to document segmentation and data mining that we hope to be able to incorporate into production during 2019. Results such as these demonstrate our commitment to R&D, the collaboration with leading research centers in their areas and the impact it has on continuing to offer a better service to our customers.