Volume 20 (2020): Issue 5 (December 2020)
Special issue on Innovations in Intelligent Systems and Applications

The volume contains extended versions of selected papers, presented at the International Symposium on INnovations in Intelligent SysTems and Applications (INISTA), held in Sofia, Bulgaria in 2019.

The main aim of this study consists of proposing a simple but effective and robust approach for PID type fuzzy controller (Fuzzy-PID) in order to improve the dynamics and stability of a magnetic ball levitation system. The design parameters of the proposed controller are optimally determined based on Cuckoo Search (CS) algorithm. During the optimization, a time domain objective function is used for minimizing the values of common step response characteristics for the optimal selection of the controller parameters. Robustness tests are performed to evaluate the performance of the proposed controller through extensive simulations under load disturbance, parametric variation and changes in references. Moreover, to show the advantage and compare the performance of the proposed controller, the PID and Fractional Order PID (FOPID) controllers tuned with CS are designed. The simulation results and comparisons with the CS based PID and FOPID controllers demonstrate that the CS based Fuzzy-PID controller has superior performance depending on small overshoot, short settling time, fast rise time and minimum steady state error. Compared with the PID and FOPID controller tuned with CS, the simulation results show that the proposed Fuzzy-PID controller tuned with CS outperforms in terms of the accuracy, robustness and the least control effort.

Traditionally, the engineers analyze signals in the time domain and in the frequency domain. These signal representations discover different signal characteristics and in many cases, the exploration of a single signal presentation is not sufficient. In the present paper, a new self-similar decomposition of digital signals is proposed. Unlike some well-known approaches, the newly proposed method for signal decomposition and description does not use pre-selected templates such as sine waves, wavelets, etc. It is realized in time domain but at the same time, it contains information about frequency signal characteristics. Good multiscale characteristics of the algorithm being proposed are demonstrated in a series of examples. It can be used for compact signal presentation, restoration of distorted signals, event detection, localization, etc. The method is also suitable for description of highly repetitive continuous and digital signals.

In this paper we present two applications of a new Belief Function-based Inter-Criteria Analysis (BF-ICrA) approach for the assessment of redundancy of criteria involved in Multi-Criteria Decision-Making (MCDM) problems. This BF-ICrA method allows to simplify the original MCDM problem by suppressing redundant criteria (if any) and thus diminish the complexity of MCDM problem. This approach is appealing for solving large MCDM problems whose solution requires the fusion of many belief functions. We show how this approach can be used in two distinct fields of applications: The GPS surveying problem, and the car selection problem.

Offline signature is one of the frequently used biometric traits in daily life and yet skilled forgeries are posing a great challenge for offline signature verification. To differentiate forgeries, a variety of research has been conducted on hand-crafted feature extraction methods until now. However, these methods have recently been set aside for automatic feature extraction methods such as Convolutional Neural Networks (CNN). Although these CNN-based algorithms often achieve satisfying results, they require either many samples in training or pre-trained network weights. Recently, Capsule Network has been proposed to model with fewer data by using the advantage of convolutional layers for automatic feature extraction. Moreover, feature representations are obtained as vectors instead of scalar activation values in CNN to keep orientation information. Since signature samples per user are limited and feature orientations in signature samples are highly informative, this paper first aims to evaluate the capability of Capsule Network for signature identification tasks on three benchmark databases. Capsule Network achieves 97 96, 94 89, 95 and 91% accuracy on CEDAR, GPDS-100 and MCYT databases for 64×64 and 32×32 resolutions, which are lower than usual, respectively. The second aim of the paper is to generalize the capability of Capsule Network concerning the verification task. Capsule Network achieves average 91, 86, and 89% accuracy on CEDAR, GPDS-100 and MCYT databases for 64×64 resolutions, respectively. Through this evaluation, the capability of Capsule Network is shown for offline verification and identification tasks.

In recent years, due to its non-volatile memory, non-locality, and weak singularity features, fractional calculations have begun to take place frequently in artificial neural network implementations and learning algorithms. Therefore, there is a need for circuit element implementations providing fractional function behaviors for the physical realization of these neural networks. In this study, a previously defined integer order memristor element equation is changed and a fractional order memristor is given in a similar structure. By using the obtained mathematical equation, frequency-dependent pinched hysteresis loops are obtained. A memristance simulator circuit that provides the proposed mathematical relationship is proposed. Spice simulations of the circuit are run and it is seen that they are in good agreement with the theory. Also, the non-volatility feature has been demonstrated with Spice simulations. The proposed circuit can be realized by using the integrated circuit elements available on the market. With a small connection change, the proposed structure can be used to produce both positive and negative memristance values.

In this work we address the problem of optimizing collective profitability in semi-competitive intermediation networks defined as augmented directed acyclic graphs. Network participants are modeled as autonomous agents endowed with private utility functions. We introduce a mathematical optimization model for defining pricing strategies of network participants. We employ welfare economics aiming to maximize the Nash social welfare of the intermediation network. The paper contains mathematical results that theoretically prove the existence of such optimal strategies. We also discuss computational results that we obtained using a nonlinear convex numerical optimization package.

The problem of thematic indexing of Open Educational Resources (OERs) is often a time-consuming and costly manual task, relying on expert knowledge. In addition, a lot of online resources may be poorly annotated with arbitrary, ad-hoc keywords instead of standard, controlled vocabularies, a fact that stretches up the search space and hampers interoperability. In this paper, we propose an approach that facilitates curators and instructors to annotate thematically educational content. To achieve this, we combine explicit knowledge graph representations with vector-based learning of formal thesaurus terms. We apply this technique in the domain of biomedical literature and show that it is possible to produce a reasonable set of thematic suggestions which exceed a certain similarity threshold. Our method yields acceptable levels for precision and recall against corpora already indexed by human experts. Ordering of recommendations is significant and this approach can also have satisfactory results for the ranking problem. However, traditional IR metrics may not be adequate due to semantic relations amongst recommended terms being underutilized.