Explainability has become one of the most discussed topics in machine learning research recent years, and although a lot methodologies that try to provide explanations black-box models have been proposed address such an issue, little discussion made on pre-processing steps involving pipeline development solutions, as feature selection. In this work, we evaluate game-theoretic approach used explain output any model, SHAP, selection mechanism. experiments, show besides being able decisions it...

Accessibility reviews collected from app stores may contain valuable information for improving apps accessibility. Recent studies have presented insightful on accessibility reviews, but they were based small datasets and focused general concerns. In this paper, we analyzed that report issues affecting users with visual disabilities or conditions. Such identified selection criteria applied over 179,519,598 of popular the Google Play Store. Our results show only 0,003% user mention conditions;...

Dimensionality reduction (DR) techniques help analysts to understand patterns in high-dimensional spaces. These techniques, often represented by scatter plots, are employed diverse science domains and facilitate similarity analysis among clusters data samples. For datasets containing many granularities or when follows the information visualization mantra, hierarchical DR most suitable approach since they present major structures beforehand details on demand. This work presents HUMAP, a novel...

Dimensionality Reduction (DR) scatterplot layouts have become a ubiquitous visualization tool for analyzing multidimensional datasets. Despite their popularity, such scatterplots suffer from occlusion, especially when informative glyphs are used to represent data instances, potentially obfuscating critical information the analysis under execution. Different strategies been devised address this issue, either producing overlap-free that lack powerful capabilities of contemporary DR techniques...

In the context of Visualization, Multidimensional Projection techniques are employed to show similarity relations among instances a multidimensional dataset. Distinct projection use different approaches perform dimensionality reduction and, consequently, metrics assess quality according and structures preservation. Usually, measures computed from whole projection, what can impair specific evaluation. This work presents novel approach evaluation on projections, in which clusters selectively...

Multidimensional projection techniques provide graphical representations computed based on instance similarities to enable the analysis of abstract and possibly large data sets. However, when set size grows these can hardly avoid overlap among markers. To overcome this issue, while some attempt remove after multidimensional projection, were developed considering non-overlapping constraints. In work, we present an four removal two non-overlapping. The evaluation was performed according five...

Dimensionality reduction (DR) techniques help analysts understand patterns in high-dimensional spaces. These techniques, often represented by scatter plots, are employed diverse science domains and facilitate similarity analysis among clusters data samples. For datasets containing many granularities or when follows the information visualization mantra, hierarchical DR most suitable approach since they present major structures beforehand details on demand. However, current not fully capable...

Dimensionality reduction is one of the most used transformations data and plays a critical role in maintaining meaningful properties while transforming from high- to low-dimensional spaces. Previous studies, e.g., on image analysis, comparing these two spaces have found that, generally, any study related anomaly detection can achieve same or similar results when applied both dimensional However, there been no studies that compare differences strategy based Kittler’s Taxonomy (ADS-KT). This...

Interpretation of machine learning models has become one the most important research topics due to necessity maintaining control and avoiding bias in these algorithms. Since many algorithms are published every day, there is a need for novel model-agnostic interpretation approaches that could be used interpret great variety Thus, advantageous way feed different input data understand changes prediction. Using such an approach, practitioners can define relations among patterns model's decision....

Dimensionality Reduction is a commonly used method to reduce the number of dimensions data. In this work, we verified its influence in classification process using combinations projection techniques as dimensionality reduction algorithms. We also Naïve Bayes and SMO classifiers.

In production lines for monitors and displays, some validation tests are based on visual inspection, whose preliminary step usually consists in detecting the area corresponding to a monitor's screen, which is then followed by evaluation procedures. Nonetheless, depending chosen technique, such detection may consume much of given test's total time. The present paper addresses mentioned problem presents an approach that uses semantic segmentation convolutional neural networks screen...

Optimum-path forest (OPF) is a graph based classifier in which the training process computes optimum-path trees rooted by prototype instances. Thus, one or more represent each class and testing on identifying tree would contain test sample. Usually, OPF performance analyzed measures computed from process, such as f-score correct classification rate (accuracy). This paper proposes an approach visualization to support understanding of processes. The visual uses multidimensional projection...

Multidimensional projection techniques have been widely used to visually explore datasets due their ability generate representations that preserve similarity relations of data points into lower dimensional spaces. To evaluate if the embedded space reflects high-dimensional structures, measures are usually employed return a quality score whole projection. In contrast this idea, we layouts by assessing each class at time using well-known measures. addition, propose multidimensional ROC curves....

Dimensionality Reduction (DR) scatterplot layouts have become a ubiquitous visualization tool for analyzing multidimensional datasets. Despite their popularity, such scatterplots suffer from occlusion, especially when informative glyphs are used to represent data instances, potentially obfuscating critical information the analysis under execution. Different strategies been devised address this issue, either producing overlap-free that lack powerful capabilities of contemporary DR techniques...