Narrow Results

We define the problem of optimizing the architecture of a multilayer perceptron (MLP) as a state space search and propose the MOST (Multiple Operators using Statistical Tests) framework that incrementally modifies the structure and checks for improvement using cross-validation. We consider five variants that implement forward/backward search, using single/multiple operators and searching depth-first/breadth-first. On 44 classification and 30 regression datasets, we exhaustively search for the optimal and evaluate the goodness based on: (1) Order, the accuracy with respect to the optimal and (2) Rank, the computational complexity. We check for the effect of two resampling methods (5 × 2, ten-fold cv), four statistical tests (5 × 2 cv t, ten-fold cv t, Wilcoxon, sign) and two corrections for multiple comparisons (Bonferroni, Holm). We also compare with Dynamic Node Creation (DNC) and Cascade Correlation (CC). Our results show that: (1) On most datasets, networks with few hidden units are optimal, (2) forward searching finds simpler architectures, (3) variants using single node additions (deletions) generally stop early and get stuck in simple (complex) networks, (4) choosing the best of multiple operators finds networks closer to the optimal, (5) MOST variants generally find simpler networks having lower or comparable error rates than DNC and CC.

As faster Random Number Generators become available, the possibility to improve the accuracy of randomness tests through the analysis of a larger number of generated bits increases. In this paper we first introduce a high-performance true-random number generator designed by authors, which use a set of discrete-time piecewise-linear chaotic maps as its entropy source. Then, we present by means of suitably improved randomness tests, the validation of this generator and the comparison with other high-end solutions. We consider the NIST test suite SP 800-22 and we show that, as suggested by NIST itself, to increase the so-called power of the test, a more in-depth analysis should be performed using the outcomes of the suite over many generated sequences. With this approach we build a framework for RNG high quality testing, with which we are able to show that the designed prototype has a comparable quality with respect to the other high-quality commercial solutions, with a working speed that is one order of magnitude faster.

Statistical testing involves the testing of software by selecting test cases from a probability distribution that is intended to represent the software's operational usage. In this paper, we describe and evaluate a framework for statistical testing of software components that incorporates test case execution and output evaluation. An operational profile and a test oracle are essential for the statistical testing of software components because they are used for test case generation and output evaluation respectively. An operational profile is a set of input events and their associated probabilities of occurrence expected in actual operation. A test oracle is a mechanism that is used to check the results of test cases. We present four types of operational profiles and three types of test oracles, and empirically evaluate them using the framework by applying them to two software components. The results show that while simple operational profiles may be effective for some components, more sophisticated profiles are needed for others. For the components that we tested, the fault-detecting effectiveness of the test oracles was similar.

In this study, we introduce a notion of stability of information granules. Granulation of information results in a series of chunks of information usually referred to as information granules. Information granules are basic building entities involved in the formation of a broad class of systems.

Information granules are percepts — entities being perceived by humans as being essential when working with some real-world phenomena, especially describing and interacting with them. The percepts need to be comprehensible. They should also reflect the experimental evidence. All in all, they should be stable meaning that they are conceptual entities that reconcile experimental reality with the subjective and ultimately observer-based judgment about the environment. Once being stable, information granules could be viewed as architecture-independent. The proposed algorithmic environment supporting this concept dwells on the ideas of statistical inference that helps quantify stability through a nonparametric testing. The χ² goodness-of-fit test is used here as a validation mechanism. First, the study elaborates on the formation of information granules and concentrates on the descriptive and prescriptive ways of their design. In the sequel, it is revealed how these two ways interact with the construction of stable information granules. A number of experimental studies are also included.