In biometrics, methods which are able to precisely adapt to the biometric features of users are much sought after. They use various methods of artificial intelligence, in particular methods from the group of soft computing. In this paper, we focus on on-line signature verification. Such signatures are complex objects described not only by the shape but also by the dynamics of the signing process. In standard devices used for signature acquisition (with an LCD touch screen) this dynamics may include pen velocity, but sometimes other types of signals are also available, e.g. pen pressure on the screen surface (e.g. in graphic tablets), the angle between the pen and the screen surface, etc. The precision of the on-line signature dynamics processing has been a motivational springboard for developing methods that use signature partitioning. Partitioning uses a well-known principle of decomposing the problem into smaller ones. In this paper, we propose a new partitioning algorithm that uses capabilities of the algorithms based on populations and fuzzy systems. Evolutionary-fuzzy partitioning eliminates the need to average dynamic waveforms in created partitions because it replaces them. Evolutionary separation of partitions results in a better matching of partitions with reference signatures, eliminates dispro-portions between the number of points describing dynamics in partitions, eliminates the impact of random values, separates partitions related to the signing stage and its dynamics (e.g. high and low velocity of signing, where high and low are imprecise-fuzzy concepts). The operation of the presented algorithm has been tested using the well-known BioSecure DS2 database of real dynamic signatures.
Content-based image retrieval methods develop rapidly with a growing scale of image repositories. They are usually based on comparing and indexing some image features. We developed a new algorithm for finding objects in images by traversing their edges. Moreover, we describe the objects by histograms of local features and angles. We use such a description to retrieve similar images fast. We performed extensive experiments on three established image datasets proving the effectiveness of the proposed method.
In this paper, we propose a new population-based evolutionary algorithm that automatically configures the used search mechanism during its operation, which consists in choosing for each individual of the population a single evolutionary operator from the pool. The pool of operators comes from various evolutionary algorithms. With this idea, a flexible balance between exploration and exploitation of the problem domain can be achieved. The approach proposed in this paper might offer an inspirational alternative in creating evolutionary algorithms and their modifications. Moreover, different strategies for mutating those parts of individuals that encode the used search operators are also taken into account. The effectiveness of the proposed algorithm has been tested using typical benchmarks used to test evolutionary algorithms.
We consider the problem of multi agents cooperating in a partially-observable environment. Agents must learn to coordinate and share relevant information to solve the tasks successfully. This article describes Asynchronous Advantage Actor-Critic with Communication (A3C2), an end-to-end differentiable approach where agents learn policies and communication protocols simultaneously. A3C2 uses a centralized learning, distributed execution paradigm, supports independent agents, dynamic team sizes, partially-observable environments, and noisy communications. We compare and show that A3C2 outperforms other state-of-the-art proposals in multiple environments.
Clustering is an attractive technique used in many fields in order to deal with large scale data. Many clustering algorithms have been proposed so far. The most popular algorithms include density-based approaches. These kinds of algorithms can identify clusters of arbitrary shapes in datasets. The most common of them is the Density-Based Spatial Clustering of Applications with Noise (DBSCAN). The original DBSCAN algorithm has been widely applied in various applications and has many different modifications. However, there is a fundamental issue of the right choice of its two input parameters, i.e the eps radius and the MinPts density threshold. The choice of these parameters is especially difficult when the density variation within clusters is significant. In this paper, a new method that determines the right values of the parameters for different kinds of clusters is proposed. This method uses detection of sharp distance increases generated by a function which computes a distance between each element of a dataset and its k-th nearest neighbor. Experimental results have been obtained for several different datasets and they confirm a very good performance of the newly proposed method.
The presented research is designed to meet a particular challenge facing the industry. Its aim is to automate the process of friction coefficient determination, using a method that enables quick and easy repeatability of measurements developed by S.C. Plasmaterm S.A in Târgu Mureş.
The Mixed CAD Generating Method, developed by the first author and presented in previous papers, is able to generate gear teeth gaps from a special points cloud. The generation method requires only a few specific points from the cutting edges of the generating tools. These points can be obtained in a first approach through a simple drawing of the cutting edges. The drawings can use either mathematical equations, or simply the construction and design principles of the cutting tools. In the case of multi-edge cutting tools of a higher level of complexity, or in case of the absence of the edge equations, there exists a simpler approach. It consists in building a solid model, or obtaining the solid model of the tools from the tool’s designer or manufacturer. In these cases, the generating points are downloaded from the solid model. This paper presents two possibilities of obtaining these points with usual CAD methods.
In this paper, we look closely at the issue of contaminated data sets, where apart from legitimate (proper) patterns we encounter erroneous patterns. In a typical scenario, the classification of a contaminated data set is always negatively influenced by garbage patterns (referred to as foreign patterns). Ideally, we would like to remove them from the data set entirely. The paper is devoted to comparison and analysis of three different models capable to perform classification of proper patterns with rejection of foreign patterns. It should be stressed that the studied models are constructed using proper patterns only, and no knowledge about the characteristics of foreign patterns is needed. The methods are illustrated with a case study of handwritten digits recognition, but the proposed approach itself is formulated in a general manner. Therefore, it can be applied to different problems. We have distinguished three structures: global, local, and embedded, all capable to eliminate foreign patterns while performing classification of proper patterns at the same time. A comparison of the proposed models shows that the embedded structure provides the best results but at the cost of a relatively high model complexity. The local architecture provides satisfying results and at the same time is relatively simple.
Today’s industry relies not only on raw material processing, but also on information. The huge amount of data obtained during the production process of goods and services; as well as information about the context of the production processes has made it necessary to account for, methodize, analyze and react in order to achieve a competitive market share. Fortunately, this technological leap has been able to support the above process, but the presence of those new technologies requires adequate preparation from the point of view of human resources too.