Logo
Logo
Data Postdiction Pipeline presented at ADBIS 2023
  • by ADMT
  • September 6, 2023

Kudos to Scott Heyman, Anna Baskin, and Brian T. Nixon for collaborating and presenting the paper, titled “Remembering the Forgotten: Clustering, Outlier Detection, and Accuracy Tuning in a Postdiction Pipeline” at the European Conference on Advances in Databases and Information Systems (ADBIS2023) in Spain. For further details, please refer to the paper under the ADMT Lab’s Publications.

The data postdiction pipeline which consists of the following components (from left to right): Input which includes the dataset and an optional error tolerance threshold, outlier detection and clustering methods on the datasets, the machine learning models that are used for postdiction, accuracy tuning module for removing data instances which violate the error threshold from the postdiction candidate pool, and output of the pipeline which includes the outlier table (data that was not decayed), a recovery table for looking up decayed values, the machine learning models, and statistics such as data sizes, clusters, etc.
  • 0 Comments
ADMT © 2020 All Right Reserved