Volume 28 (2018): Edizione 4 (December 2018)

In this paper, for multiple different chaotic systems with fully unknown parameters, a novel synchronization scheme called ‘modified function projective multi-lag generalized compound synchronization’ is put forward. As an advantage of the new method, not only the addition and subtraction, but also the multiplication of multiple chaotic systems are taken into consideration. This makes the signal hidden channels more abundant and the signal hidden methods more flexible. By virtue of finite-time stability theory and an adaptive control technique, a finite-time adaptive control scheme is established to realize the finite-time synchronization and to properly evaluate the unknown parameters. A detailed theoretical derivation and a specific numerical simulation demonstrate the feasibility and validity of the advanced scheme.

The problem of practical synchronization of an uncertain Duffing oscillator with a higher order chaotic system is considered. Adaptive control techniques are used to obtain chaos synchronization in the presence of unknown parameters and bounded, unstructured, external disturbances. The features of the proposed controllers are compared by solving Duffing-Arneodo and Duffing-Chua synchronization problems.

The problem of fault diagnosis in hybrid systems is investigated. It is assumed that the hybrid systems under consideration consist of a finite automaton, a set of nonlinear difference equations and the so-called mode activator that coordinates the action of the other two parts. To solve the fault diagnosis problem, hybrid residual generators based on both diagnostic observers and parity relations are used. It is shown that the hybrid nature of the system imposes some restrictions on the possibility of creating such generators. Sufficient solvability conditions of the fault diagnosis problem are found. Examples illustrate details of the solution.

A new, state space, discrete-time, and memory-efficient model of a one-dimensional heat transfer process is proposed. The model is derived directly from a time-continuous, state-space semigroup one. Its discrete version is obtained via a continuous fraction expansion method applied to the solution of the state equation. Fundamental properties of the proposed model, such as decomposition, stability, accuracy and convergence, are also discussed. Results of experiments show that the model yields good accuracy in the sense of the mean square error, and its size is significantly smaller than that of the model employing the well-known power series expansion approximation.

In this paper we present interconnection and damping assignment passivity-based control (IDA-PBC) applied to a 2 degrees of freedom (DOFs) underactuated gyroscope. First, the equations of motion of the complete system (3-DOF) are presented in both Lagrangian and Hamiltonian formalisms. Moreover, the conditions to reduce the system from a 3-DOF to a 2- DOF gyroscope, by using Routh’s equations of motion, are shown. Next, the solutions of the partial differential equations involved in getting the proper controller are presented using a reduction method to handle them as ordinary differential equations. Besides, since the gyroscope has no potential energy, it presents the inconvenience that neither the desired potential energy function nor the desired Hamiltonian function has an isolated minimum, both being only positive semidefinite functions; however, by focusing on an open-loop nonholonomic constraint, it is possible to get the Hamiltonian of the closed-loop system as a positive definite function. Then, the Lyapunov direct method is used, in order to assure stability. Finally, by invoking LaSalle’s theorem, we arrive at the asymptotic stability of the desired equilibrium point. Experiments with an underactuated gyroscopic mechanical system show the effectiveness of the proposed scheme.

Designing a tracking control system for an over-actuated dynamic positioning marine vessel in the case of insufficient information on environmental disturbances, hydrodynamic damping, Coriolis forces and vessel inertia characteristics is considered. The designed adaptive MIMO backstepping control law with control allocation is based on Lyapunov control theory for cascaded systems to guarantee stabilization of the marine vessel position and heading. Forces and torque computed from the adaptive control law are allocated to individual thrusters by employing the quadratic programming method in combination with the cascaded generalized inverse algorithm, the weighted least squares algorithm and the minimal least squares algorithm. The effectiveness of the proposed control scheme is demonstrated by simulations involving a redundant set of actuators. The evaluation criteria include energy consumption, robustness, as well accuracy of tracking during typical vessel operation.

We consider the discrete-time G/GI/1 queueing system with multiple exhaustive vacations. By a transform approach, we obtain an expression for the probability generating function of the waiting time of customers in such a system. We then show that the results can be used to assess the performance of G/GI/1 queueing systems with server breakdowns as well as that of the low-priority queue of a preemptive M^X+G/GI/1 priority queueing system. By calculating service completion times of low-priority customers, various preemptive breakdown/priority disciplines can be studied, including preemptive resume and preemptive repeat, as well as their combinations. We illustrate our approach with some numerical examples.

We propose new methods for support vector machines using a tree architecture for multi-class classification. In each node of the tree, we select an appropriate binary classifier, using entropy and generalization error estimation, then group the examples into positive and negative classes based on the selected classifier, and train a new classifier for use in the classification phase. The proposed methods can work in time complexity between O(log₂ N) and O(N), where N is the number of classes. We compare the performance of our methods with traditional techniques on the UCI machine learning repository using 10-fold cross-validation. The experimental results show that the methods are very useful for problems that need fast classification time or those with a large number of classes, since the proposed methods run much faster than the traditional techniques but still provide comparable accuracy.

Radial basis function networks (RBFNs) or extreme learning machines (ELMs) can be seen as linear combinations of kernel functions (hidden neurons). Kernels can be constructed in random processes like in ELMs, or the positions of kernels can be initialized by a random subset of training vectors, or kernels can be constructed in a (sub-)learning process (sometimes by k-means, for example). We found that kernels constructed using prototype selection algorithms provide very accurate and stable solutions. What is more, prototype selection algorithms automatically choose not only the placement of prototypes, but also their number. Thanks to this advantage, it is no longer necessary to estimate the number of kernels with time-consuming multiple train-test procedures. The best results of learning can be obtained by pseudo-inverse learning with a singular value decomposition (SVD) algorithm. The article presents a comparison of several prototype selection algorithms co-working with singular value decomposition-based learning. The presented comparison clearly shows that the combination of prototype selection and SVD learning of a neural network is significantly better than a random selection of kernels for the RBFN or the ELM, the support vector machine or the kNN. Moreover, the presented learning scheme requires no parameters except for the width of the Gaussian kernel.

Due to the advances made in recent years, methods based on deep neural networks have been able to achieve a state-of-the-art performance in various computer vision problems. In some tasks, such as image recognition, neural-based approaches have even been able to surpass human performance. However, the benchmarks on which neural networks achieve these impressive results usually consist of fairly high quality data. On the other hand, in practical applications we are often faced with images of low quality, affected by factors such as low resolution, presence of noise or a small dynamic range. It is unclear how resilient deep neural networks are to the presence of such factors. In this paper we experimentally evaluate the impact of low resolution on the classification accuracy of several notable neural architectures of recent years. Furthermore, we examine the possibility of improving neural networks’ performance in the task of low resolution image recognition by applying super-resolution prior to classification. The results of our experiments indicate that contemporary neural architectures remain significantly affected by low image resolution. By applying super-resolution prior to classification we were able to alleviate this issue to a large extent as long as the resolution of the images did not decrease too severely. However, in the case of very low resolution images the classification accuracy remained considerably affected.

Personal identification is particularly important in information security. There are numerous advantages of using electroencephalogram (EEG) signals for personal identification, such as uniqueness and anti-deceptiveness. Currently, many researchers focus on single-dataset personal identification, instead of the cross-dataset. In this paper, we propose a method for cross-dataset personal identification based on a brain network of EEG signals. First, brain functional networks are constructed from the phase synchronization values between EEG channels. Then, some attributes of the brain networks including the degree of a node, the clustering coefficient and global efficiency are computed to form a new feature vector. Lastly, we utilize linear discriminant analysis (LDA) to classify the extracted features for personal identification. The performance of the method is quantitatively evaluated on four datasets involving different cognitive tasks: (i) a four-class motor imagery task dataset in BCI Competition IV (2008), (ii) a two-class motor imagery dataset in the BNCI Horizon 2020 project, (iii) a neuromarketing dataset recorded by our laboratory, (iv) a fatigue driving dataset recorded by our laboratory. Empirical results of this paper show that the average identification accuracy of each data set was higher than 0.95 and the best one achieved was 0.99, indicating a promising application in personal identification.

Modern cancer diagnostics is based heavily on cytological examinations. Unfortunately, visual inspection of cytological preparations under the microscope is a tedious and time-consuming process. Moreover, intra- and inter-observer variations in cytological diagnosis are substantial. Cytological diagnostics can be facilitated and objectified by using automatic image analysis and machine learning methods. Computerized systems usually preprocess cytological images, segment and detect nuclei, extract and select features, and finally classify the sample. In spite of the fact that a lot of different computerized methods and systems have already been proposed for cytology, they are still not routinely used because there is a need for improvement in their accuracy. This contribution focuses on computerized breast cancer classification. The task at hand is to classify cellular samples coming from fine-needle biopsy as either benign or malignant. For this purpose, we compare 5 methods of nuclei segmentation and detection, 4 methods of feature selection and 4 methods of classification. Nuclei detection and segmentation methods are compared with respect to recall and the F1 score based on the Jaccard index. Feature selection and classification methods are compared with respect to classification accuracy. Nevertheless, the main contribution of our study is to determine which features of nuclei indicate reliably the type of cancer. We also check whether the quality of nuclei segmentation/detection significantly affects the accuracy of cancer classification. It is verified using the test set that the average accuracy of cancer classification is around 76%. Spearman’s correlation and chi-square test allow us to determine significantly better features than the feature forward selection method.

The paper presents a novel spectral algorithm EVSA (eigenvector structure analysis), which uses eigenvalues and eigenvectors of the adjacency matrix in order to discover clusters. Based on matrix perturbation theory and properties of graph spectra we show that the adjacency matrix can be more suitable for partitioning than other Laplacian matrices. The main problem concerning the use of the adjacency matrix is the selection of the appropriate eigenvectors. We thus propose an approach based on analysis of the adjacency matrix spectrum and eigenvector pairwise correlations. Formulated rules and heuristics allow choosing the right eigenvectors representing clusters, i.e., automatically establishing the number of groups. The algorithm requires only one parameter-the number of nearest neighbors. Unlike many other spectral methods, our solution does not need an additional clustering algorithm for final partitioning. We evaluate the proposed approach using real-world datasets of different sizes. Its performance is competitive to other both standard and new solutions, which require the number of clusters to be given as an input parameter.

Despite the rapid growth of other types of social media, Internet discussion forums remain a highly popular communication channel and a useful source of text data for analyzing user interests and sentiments. Being suited to richer, deeper, and longer discussions than microblogging services, they particularly well reflect topics of long-term, persisting involvement and areas of specialized knowledge or experience. Discovering and characterizing such topics and areas by text mining algorithms is therefore an interesting and useful research direction. This work presents a case study in which selected classification algorithms are applied to posts from a Polish discussion forum devoted to psychoactive substances received from home-grown plants, such as hashish or marijuana. The utility of two different vector text representations is examined: the simple bag of words representation and the more refined embedded global vectors one. While the former is found to work well for the multinomial naive Bayes algorithm, the latter turns out more useful for other classification algorithms: logistic regression, SVMs, and random forests. The obtained results suggest that post-classification can be applied for measuring publication intensity of particular topics and, in the case of forums related to psychoactive substances, for monitoring the risk of drug-related crime.

In the era of big data, solutions are desired that would be capable of efficient data reduction. This paper presents a summary of research on an algorithm for complementation of a Boolean function which is fundamental for logic synthesis and data mining. Successively, the existing problems and their proposed solutions are examined, including the analysis of current implementations of the algorithm. Then, methods to speed up the computation process and efficient parallel implementation of the algorithm are shown; they include optimization of data representation, recursive decomposition, merging, and removal of redundant data. Besides the discussion of computational complexity, the paper compares the processing times of the proposed solution with those for the well-known analysis and data mining systems. Although the presented idea is focused on searching for all possible solutions, it can be restricted to finding just those of the smallest size. Both approaches are of great application potential, including proving mathematical theorems, logic synthesis, especially index generation functions, or data processing and mining such as feature selection, data discretization, rule generation, etc. The problem considered is NP-hard, and it is easy to point to examples that are not solvable within the expected amount of time. However, the solution allows the barrier of computations to be moved one step further. For example, the unique algorithm can calculate, as the only one at the moment, all minimal sets of features for few standard benchmarks. Unlike many existing methods, the algorithm additionally works with undetermined values. The result of this research is an easily extendable experimental software that is the fastest among the tested solutions and the data mining systems.