pebblous

The performance of deep learning AI is determined by model design and dataset quality at once. Pebblous focuses more on the data side operations with <Data Clinic> solution, which is composed of three key modules such as data imaging, data diagnoisis, and data enhancement. Our <Data Imaging> module transforms a large high-dimensional dataset into a proper form with more managable and observable dimensions. Once a data image is generated, we perform <Data Diagnoisis> to assess the data quality with various mathematical and semantic measures. Then, we generate <Data Quality Report> to promote customers’ data insight and further enhancement actions. As a final step, we apply <Data Enhancement> module based on the qulity report. <Data Synthesis> is one of the such enhancement actions to fill in the missing data or to anonymize privacy data on the data manifold defined in the data imaging stage.