Performing data mining tasks such as clustering can be very complex due to the high dimensionality and volume of data being mined. This paper proposes an approach for data clustering using the Symbiotic Organisms Sear...
详细信息
ISBN:
(纸本)9788576693178
Performing data mining tasks such as clustering can be very complex due to the high dimensionality and volume of data being mined. This paper proposes an approach for data clustering using the Symbiotic Organisms Search algorithm (SOS) developed in the MapReduce parallel architecture. Also, the cluster quality evolution is analysed using the purity measured considering four different fitness metrics. The cluster qualities obtained by the proposed approach not only shows to be competitive with other approaches but also increased its performance using the MapReduce architecture. Another contribution of this work is to bring to light the correlation between the cluster purity and the fitness value obtained during the optimization process. It was noticed that for some fitness metrics the final purity found by the optimization algorithm is less than the purity found in an earlier stage in the optimization process.
Artificial bee colony(ABC) algorithm is one of bio-inspired algorithms(BIAs) that has created interest among optimization researchers because of its many successes in solving various optimization problems. Nevertheles...
详细信息
Artificial bee colony(ABC) algorithm is one of bio-inspired algorithms(BIAs) that has created interest among optimization researchers because of its many successes in solving various optimization problems. Nevertheless, it has been identified to have few limitations which are slow convergence speed and premature convergence tendency. In order to eradicate the limitations, researchers have proposed variants of ABC which have also been found to be insufficient enough to overcome the problems simultaneously. Thus, a new ABC variant has been proposed in this work to solve the stated problems. The proposed algorithm is referred to as JA-ABC2. It has shown superior results at improving the performance of ABC algorithm.
Ant Colony Optimization algorithms are the most successful and widely accepted algorithmic techniques based on the decentralized collaborative behavior of real ants when foraging for food. The initialization of pherom...
详细信息
Ant Colony Optimization algorithms are the most successful and widely accepted algorithmic techniques based on the decentralized collaborative behavior of real ants when foraging for food. The initialization of pheromone in these algorithms is an important step because it dictates the speed of the system's convergence tothe optimal solution. All the proposed initialization techniques in the literature use a single value toinitialize the pheromone on all edges. In our paper, instead of using a constant or a pre-calculated value toinitialize the pheromone the edges, we propose a local pheromone initialization technique thatinvolves the ants initializing the edges, using local information, as they encounter the edges for the first time. We tested our proposed local initialization using the Ant Colony System algorithm tosolve the Travelling Salesman Problem. Our approach, when compared tothe standard initialization approaches, provided better results in more than 70% of the tested datasets. Also, our algorithm did not require an initialization for all edges. In general, our local pheromone initialization approach was successful in achieving a balance between the solution quality and the time required toconstruct that solution even in the cases in which it was not able tofind the optimal path.
In this study a new Bat Algorithm (BA) based on multi-swarm technique called the Multi-Swarm Bat Algorithm (MSBA) is proposed to address the problem of premature convergence phenomenon. The problem happens when search...
详细信息
This paper presents new aspects of using biometric features and patterns for creation of new cryptographic protocols. These protocols are dedicated for secure information sharing and distribution processes. Such algor...
详细信息
ISBN:
(纸本)9781479985630
This paper presents new aspects of using biometric features and patterns for creation of new cryptographic protocols. These protocols are dedicated for secure information sharing and distribution processes. Such algorithms allow protecting confidential data sets, and performing secure operation for different kinds of information. bio-inspired techniques for securing information will be presented on the basis of using the selected biometrics - the voice personal features. Development of such algorithms will enable reconstruction of secured data sets exclusively by encoding and authorized parties, which received accessing grants for secret data reconstruction.
Many optimization techniques have been proposed for solving economic load and or emission dispatch problems in power generation plants. Among them, bio-inspired techniques have become popular due to their performance ...
详细信息
Many optimization techniques have been proposed for solving economic load and or emission dispatch problems in power generation plants. Among them, bio-inspired techniques have become popular due to their performance and simplicity. Bat Algorithm (BA) is one of the newer additions in the field although it has been applied in a series of evaluation tests. The method is trying to replicate the echolocation behavior of bats with artificial ones. In this paper, we apply BA to the well-known problem of economic load dispatch, in different power demand scenarios. Its performance in terms of solution fitness is compared with other optimizations techniques proposed in the literature for the same scenarios.
暂无评论