Caetano Traina

Spatial join selectivity using power laws

We discovered a surprising law governing the spatial join selectivity across two sets of points. An example of such a spatial join is “find the libraries that are within 10 miles of schools”. Our law dictates that the number of such qualifying pairs follows a power law, whose exponent we call “pair-count exponent” (PC). We show that this law also holds for self-spatial-joins (“find schools within 5 miles of other schools”) in addition to the general case that the two point-sets are distinct. Our law holds for many real datasets, including diverse environments (geographic datasets, feature vectors from biology data, galaxy data from astronomy).nIn addition, we introduce the concept of the Box-Occupancy-Product-Sum (BOPS) plot, and we show that it can compute the pair-count exponent in a timely manner, reducing the run time by orders of magnitude, from quadratic to linear. Due to the pair-count exponent and our analysis (Law 1), we can achieve accurate selectivity estimates in constant time (O(1)) without the need for sampling or other expensive operations. The relative error in selectivity is about 30% with our fast BOPS method, and even better (about 10%), if we use the slower, quadratic method.

Smart histogram analysis applied to the skull-stripping problem in T1-weighted MRI

In this paper we address the skull-stripping problem in 3D MR images. We propose a new method that employs an efficient and unique histogram analysis. A fundamental component of this analysis is an algorithm for partitioning a histogram based on the position of the maximum deviation from a Gaussian fit. In our experiments we use a comprehensive image database, including both synthetic and real MRI, and compare our method with other two well-known methods, namely BSE and BET. For all datasets we achieved superior results. Our method is also highly independent of parameter tuning and very robust across considerable variations of noise ratio.

The spatial-perceptual design space: a new comprehension for data visualization

We revisit the design space of visualizations aiming at identifying and relating its components. In this sense, we establish a model to examine the process through which visualizations become expressive for users. This model has lead us to a taxonomy oriented to the human visual perception. The essence of this taxonomy provides natural criteria in order to delineate a novel understanding for the design space of visualizations. From such understanding, we elaborate a model for generalized design. The model poses an intuitive comprehension for the visualization design space departing from fundamental pre-attentive stimuli and from perceptual phenomena. The paper is presented as a survey, its structure introduces an alternative conceptual organization for the space of techniques concerning visual analysis.

A new feature descriptor derived from Hilbert space-filling curve to assist breast cancer classification

Malignant breast tumors and benign masses appear in mammograms with different shape characteristics: the former usually have rough, spiculated, or microlobulated contours, whereas the latter commonly have smooth, round, oval, or macrolobulated contours. Visual features that characterize shape roughness can assist in distinguishing between malignant tumors and benign masses in mammo-grams. Here we propose a new approach based on Hilbert curves to classify breast masses as benign or malignant. The feature extraction is performed in linear time, and is amenable to parallel processing, whereas the classification phase can be performed by a classical neural networks structure. We evaluated our method using a set of 111 contours from 65 benign masses and 46 malignant tumors. As the experimental evaluations show, we achieved an accuracy of 0.99 in terms of the area under the receiver operating characteristics curve, a very high classification result.

The Visual Expression Process: Bridging Vision and Data Visualization

Visual data analysis follows a sequence of steps derived from perceptual faculties that emanate from the human vision system. Firstly, pre-attentive phenomena determine a map of potential interesting objectives. Then, attentive selection concentrates on one element of a vocabulary of visual perceptions. Lastly, perceptions in working memory combine to long-term domain knowledge to support cognition. Following this process, we present a model that joins vision theory and visual data analysis aiming at settling a comprehension of why graphical presentations expand the human intellect, making us smarter.

Spectral analysis and text processing over the computer science literature: patterns and discoveries

We defend the thesis that the use of text analytics can boost the results of analyses based on Singular Value Decomposition (SVD). To demonstrate our supposition, first we model the Digital Bibliography & Library Project (DBLP) as a relational schema; over this schema we use text analytics applied to the terms extracted from the titles of the articles. Then, we apply SVD on the relationships defined between these terms, publication vehicles, and authors; accordingly, we were able to identify the more representative communities and the more active authors relating them to the most meaningful terms and topics found in their respective publications. The results were semantically dense and concise, also leading to performance gains.

Mining Patterns for Visual Interpretation in a Multiple-Views Environment

This chapter introduces a novel systematization aiming at extending the application range of Information Visualization and Visual Data Mining. We present an innovative framework named Visualization Tree in order to integrate multiple data visualizations assisted by novel visual exploration techniques. These exploration techniques are named Frequency Plot, Relevance Plot and Representative Plot, and are integrated according the proposed Visualization Tree framework. The systematization of visualization techniques enabled by these concepts defines a Visual Data Mining environment where multiple presentation workspaces are kept together, linked according to analytical decisions taken by the user. Our emphasis is on developing an intuitive and versatile multiple-views system that helps the user to identify visual patterns while interpreting multiple data subsets. In this context, the analyst is able to draw and summarize several subsets that are inspected simultaneously each in a dedicated workspace.