Demos

Descripción: IQA Demonstrator (Imputation Quality Assessment). In the field of machine learning (ML), it is common to encounter variables with missing fields, meaning some observations do not have all values for all variables. The Imputation Quality Assessment (IQA) method compares various imputation methods and results in an imputation score that considers both the percentage of missing values. It is common to encounter variables with missing values, meaning some observations do not have all values for all variables.

Descripción: The EAPP method (Extended APriori Probability) helps evaluate the quality of a model trained from a given dataset. The EAPP value represents the performance of an unsupervised algorithm. Models that do not surpass this value are considered to be of low quality. On the other hand, high EAPP index values may indicate the presence of bias in the data.

Descripción: Provides explanations about predictions. Among other things, it offers data on stability, uncertainty, and graphs that show the result of the prediction when altering the variables.

Descripción: Optimal selection of points based on silhouette reconstruction from multiple views.

Descripción: Generation of images using generative AI. It uses the DreamBooth technique.

Descripción: This tool offers functionality to load datasets and train and evaluate a set of models for Anomaly Detection focused on images, available in the Surface Anomaly Detection libraries ITI-II3D.

Descripción: ITI-Gen Light is an AI model validation tool that can work offline. Queries are made through tabular data tables, either by importing an Excel/CSV file or by directly filling out the table. It allows users to view previously made queries and export those of interest to Excel. It also allows tagging each query, particularly with the actual prediction value the model was supposed to make to validate its performance. This demonstrator is associated with a classification model (case/control) of three variables: v1, a numerical value in [-1,1]; v2, an ordinal value; and v3, a categorical value.

Descripción: The tool enables segmentation of the breast, dense tissue, and the risk of developing cancer.