An open-source low-cost wireless sensor system for acquisition of human movement data.

Several fields of research such as medicine, robotics, sports, informatics, etc., require the analysis of human movement. Traditional systems for acquisition and analysis of human movement data are based on video cameras or active sensors. However, those systems are limited to high-resource settings. Wearable devices allow monitoring subjects outside typical clinical or research environments. Here, we present an open source low-cost wireless sensor system for acquisition of human movement data. Our system consists of two main parts: a server that stores data and, one or more wearable sensor modules that collect movement data through Inertial Measurement Units (IMUs) and transmit them wirelessly to the server. As a proof of concept, we measured human gait activity. Our results show that our system with IMUs can acquire quantifiable movement data. Characteristics such as open source code and its low-cost, make our system a viable alternative for clinical or research.

Analixity: An open source, low-cost analysis system for the elevated plus maze test, based on computer vision techniques.

Manual analysis of behavioral tests in rodents involves inspection of video recordings by a researcher that assesses rodent movements to quantify parameters related with a behavior of interest. The assessment of the researcher during the quantification of such parameters can introduce variability among experimental conditions or among sessions of analysis. Here, we introduce Analixity, a video processing software for the elevated plus maze test (EPM), in which quantification of behavioral parameters is automatic, reducing the time spent in analysis and solving the variability problem. Analixity is an adaptable multiplatform open-source system. Analixity generates an Excel file with the quantified behavioral variables, such as time spent in open and closed arms and in the center zone, number of entries to each zone and total distance traveled during the test. For validation, we compared results obtained by Analixity with results obtained by manual analysis. We did not find statistically significant differences. In addition, we compared the results obtained by Analixity with results obtained by the commercial software ANY-maze. We did not find statistically significant differences in the quantification of parameters such as time spent in open arms, time spent in closed arms, time spent in center zone, number of closed arms, open arms entries, and anxiety index. We concluded that Analixity is an open-source software as reliable and effective as a commercial software.

How good is crude MDL for solving the bias-variance dilemma? An empirical investigation based on Bayesian networks.

The bias-variance dilemma is a well-known and important problem in Machine Learning. It basically relates the generalization capability (goodness of fit) of a learning method to its corresponding complexity. When we have enough data at hand, it is possible to use these data in such a way so as to minimize overfitting (the risk of selecting a complex model that generalizes poorly). Unfortunately, there are many situations where we simply do not have this required amount of data. Thus, we need to find methods capable of efficiently exploiting the available data while avoiding overfitting. Different metrics have been proposed to achieve this goal: the Minimum Description Length principle (MDL), Akaike's Information Criterion (AIC) and Bayesian Information Criterion (BIC), among others. In this paper, we focus on crude MDL and empirically evaluate its performance in selecting models with a good balance between goodness of fit and complexity: the so-called bias-variance dilemma, decomposition or tradeoff. Although the graphical interaction between these dimensions (bias and variance) is ubiquitous in the Machine Learning literature, few works present experimental evidence to recover such interaction. In our experiments, we argue that the resulting graphs allow us to gain insights that are difficult to unveil otherwise: that crude MDL naturally selects balanced models in terms of bias-variance, which not necessarily need be the gold-standard ones. We carry out these experiments using a specific model: a Bayesian network. In spite of these motivating results, we also should not overlook three other components that may significantly affect the final model selection: the search procedure, the noise rate and the sample size.