AI-driven control of counting machines

AI-driven control of counting machines is the content of nugget 15, developed by the Czech Technical University of Prague.

The learning outcomes of this nugget are: use of k-means clustering algorithm to find significant partitions within data, data pre-processing, data filtering and data standardization and visualization of high-dimensional data in 2D/3D.

A counting machine is described in more detail and the connected knowledge base. The problem to solve is to automate change-over process with minimum intervention of operators, reducing workload. The approach is to use collective expert knowledge, where the biggest challenge is to handle a variety of collected data. Data pre-processing (data filtering and data standardization) is explained and also K-means clustering, which is a representative of quantization techniques that try to identify the density of large and high-dimensional data. Finally, the visualization of clusters is explained.