Recent work has unveiled a theory for reasoning about the decisions made by binary classifiers: a classifier describes a Boolean function, and the reasons behind an instance being classified as positive are the prime-implicants of the function that are satisfied by the instance. One drawback of these works is that they do not explicitly treat scenarios where the underlying data is known to be constrained, e.g., certain combinations of features may not exist, may not be observable, or may be required to be disregarded. We propose a more general theory, also based on prime-implicants, tailored to taking constraints into account. The main idea is to view classifiers as describing partial Boolean functions that are undefined on instances that do not satisfy the constraints. We prove that this simple idea results in more parsimonious reasons. That is, not taking constraints into account (e.g., ignoring, or taking them as negative instances) results in reasons that are subsumed by reasons that do take constraints into account. We illustrate this improved succinctness on synthetic classifiers and classifiers learnt from real data.

Parkinson's Disease (PD) is a complex neurodegenerative disorder that is challenging to diagnose. Recent research has demonstrated predictive value in the analysis of dynamic handwriting features for detecting PD, however, consensus on clinicallyuseful features is yet to be reached. Here we explore and evaluate secondary kinematic handwriting features hypothesized to be diagnostically relevant to Parkinson's Disease using a publicly-available Spiral Drawing Test PD dataset. Univariate and multivariate analysis was performed on derived features. Classification outcome was determined using logistic regression models with 10-fold cross validation. Feature correlation was based on model specificity and sensitivity. Variations in grip angle, instantaneous acceleration and pressure indices were found to have high predictive potential as clinical markers of PD, with combined classification accuracy of above 90%. Our results show that the significance of secondary handwriting features and recommend the feature expansion step for hypothesis generation, comparative evaluation of test types and improved classification accuracy.

In this paper we review the application of logic synthesis methods for uncovering minimal structures in observational/medical datasets. Traditionally used in digital circuit design, logic synthesis has taken major strides in the past few decades and forms the foundation of some of the most powerful concepts in computer science and data mining. Here we provide a review of current state of research in application of logic synthesis methods for data analysis and provide a demonstrative example for systematic application and reasoning based on these methods.

No abstract

Despite the widespread use of techniques and tools for causal analysis, existing methodologies still fall short as they largely regard causal variables as independent elements, thereby failing to appreciate the significance of the interactions of causal variables. The prospect of inferring causal relationships from weaker structural assumptions compels for further research in this area. This study explores the effects of the interactions of variables in the context of causal analysis, and introduces new advancements to this area of research. In this study, we introduce a new approach for the causal complexity with the goal of making the solution set closer to deterministic by taking into consideration the underlying patterns embedded within a dataset; in particular, the interactions of causal variables. Our model follows the configurational approach, and as such, is able to account for the three major phenomena of conjunctural causation, equifinality, and causal asymmetry.