Neural models based on unsupervised learning for the visualization of large data sets

New and innovative aspects

There are other data visualization methods, but most of them do not providea simple and intuitive way to identify the structure and analyze the behavior of thedata.
In this way, analysis times are reduced and preliminary results are obtained thatthey can guide later, more in-depth analyzes.

Main advantages of its use

By offering a projection that preserves most of the information in the data, maximizing the possibility of enhancing the structure of the data, a subsequent rapid analysis of the data is allowed. In the first place, the general structure of the data set can be identified, detecting their grouping (different types of customers, product groups, behavior of the analyzed machines, stages or regions of operation of the different processes ...). Once this general structure is known, it is possible to proceed to a detailed analysis of each of these groups, considering their internal structure. With all this, you can easily carry out an analysis of the data, drawing conclusions about them when interpreted by an expert in the field (factors that affect customer behavior, star products or with less success, optimal values adjustment of a certain machine, understanding of certain processes).

Specifications

Currently there is such a large amount of information that it cannot be easily processed to extract knowledge from it. An exhaustive analysis, on a case-by-case basis, would push the analysis of a large number of data that is captured today to unapproachable limits. These can refer to statistics of purchases of our clients, characteristics of different products, records of a certain machine or system, samples of a certain industrial process, etc. The projectionist models propose a data visualization with these characteristics; a large volume and also a high dimensionality (large amount of information for each of the customers, products, records, samples ...).

Applications

These models can be applied to any data set that needs to beanalyzed, regardless of the reality to which these data refer.Some application examples are those mentioned above: economic data, production data, operating or access statistics, etc.

Current development status

Device already developed and validated for industrialization.

Desired business relationship

Commercial Agreement, License Agreement, Technical Cooperation: further development; Technical Cooperation: testing new applications; Technical Cooperation: adaptation to specific needs.