With the development of bio-computing technology, the research for DNA watermarking, which considers DNA information a medium, has been showing interest. In particular, there is a need for a reversible DNA watermarking technology capable of DNA storage and forgery prevention of DNA sequence, and analyzing biological mutation processes by watermark while recovering perfectly the original DNA sequence. In this paper, we address a reversible watermarking method for noncoding DNA sequences using an adaptive prediction error expansion based on least square predictor. In our method, the 4-character nucleotide sequences of the noncoding region are converted into code values by the adjacent n nucleotide bases. Then, a least squares based prediction error for the current code coefficient is obtained, and this prediction error is expanded adaptively by the number of bits determined according to the condition of prediction error expansion. Here, a false start codon generation is prevented through a comparison search between the watermarked adjacent base sequences. The experimental results showed that our method has a higher watermark capacity than the conventional method and the mean prediction error extension method, and does not generate biological mutations and false start codons.
Deep learning is considered to be a model-free, end-to-end, and black-box approach. It requires numerous data samples instead of expert knowledge on the target domain. Hence, it does not specify the mechanism and reasons for its decision making. This aspect is considered a critical limitation of deep learning. This paper introduces another viewpoint, namely Bayesian deep learning. Deep learning can be installed in any framework, such as Bayesian networks and reinforcement learning. Subsequently, an expert can implement the knowledge as the graph structure, accelerate learning, and obtain new knowledge on the target domain. The framework is termed as the deep generative model. Conversely, we can directly introduce the Bayesian modeling approach to deep learning. Subsequently, it is possible to explore deep learning with respect to the confidence of its decision making via uncertainty quantification of the output and detect wrong decision-making or anomalous inputs. Given the aforementioned approaches, it is possible to adjust the “brightness” of deep learning.
Liquid State Machine (LSM) is one of reservoir computing models and due to universal computing capability its application and improvement attract researchers. Meanwhile development of measurement technology reveals the existence of specific structure in brain networks, such as scale-free, small-world, and modular properties, which contribute to higher-order brain functions such as cognition and memory. In this paper, we apply various network models to a recurrent neural network of an LSM and investigate the relationship between structural properties and accuracy of discrimination. Results suggest that modularity of a recurrent neural network enhances discrimination capability of LSM.
The structure of social networks or human relationships is difficult to understand since we cannot observed their links and link weights directly. The network resonance method was proposed to obtain information on the unknown Laplacian matrix representing the social network structure. This method extracts information on the eigenvalues and eigenvectors of the Laplacian matrix by observing user dynamics on social networks. The original Laplacian matrix can be reconstructed if all eigenvalues and eigenvectors are known. However, the network resonance method has a problem: the information available about eigenvectors is limited to the absolute value of each element. Therefore, to determine the Laplacian matrix, it is necessary to determine the signs of each element of all eigenvectors. However, sign determination incurs the computation cost of the order of O(2n) for each of n eigenvectors. This paper proposes a method to determine the signs of each eigenvector element efficiently. The main idea of the method is to generate n2-n different sign determination problems for n eigenvectors and to solve them in parallel. All that is required is to obtain n different eigenvectors determined in the shortest time from the n2-n different sign determination problems. Since the ratio of the number of sign determinations completed in the method is 1/(n-1), its efficiency rises with the number of network users. In addition, simulations on networks generated by the BA model show that proposed method offers sign determination in polynomial time.
Mobile wireless communication demand has increased rapidly in recent years, and advanced mobile devices such as smartphones have been widely deployed. Those devices are generally capable of connecting various wireless networks such as cellular networks and wireless LANs. Due to the software and hardware constraint of mobile devices, it is necessary to develop efficient and light-weight algorithms to select wireless networks. Previous studies have shown that multi-armed bandit algorithms can efficiently select wireless channel in cognitive radio. In this paper, we propose an efficient wireless network selection technique by using a multi-armed bandit algorithm called tug-of-war (TOW). We implement the proposed algorithm to a wireless device and show the effectiveness of the proposed method by experimental demonstrations.
In order to invigorate utilization of high-performance computing (HPC) in industries, we have described the business building block canvas and reconsidered on HPC service model that HPC service providers would take a strategic planning to give their own abilities to small industries. And we have taken a restructuring process of the business building blocks canvas to make an innovative service strategy that HPC service organizations would adopt as their own user service model. The economic benefits of the small industries through adopting HPC services provided from the HPC service organizations effectively and efficiently have increased when the HPC service model is proportionate, targeted to the industrial present problems and introduced only when absolutely necessary. We have described two differentiated HPC service models and their achievements, and examined how effective the HPC service model of the HPC service organizations is composed to provide HPC abilities to the industry. The approach of the business building blocks canvas would help the HPC organizations to analyze and restructure the strategy.
The Steiner tree problem in graphs is an NP-hard combinatorial optimization problem. To solve the NP-hard combinatorial optimization problems, such as the traveling salesman problems, the quadratic assignment problems, and the vehicle routing problems, an algorithm for searching solutions by chaotic dynamics, or the chaotic search, exhibits good performance. From this viewpoint, this paper proposes an algorithm for solving the Steiner tree problem in graphs with chaotic dynamics. Comparing the performance of the chaotic search with that of the tabu search, we analyze the searching characteristics of the chaotic search. From the results of numerical experiments, we found that if parameters of the chaotic search are set to appropriate values, the chaotic search exhibits good performance and the chaotic search outputs optimal solutions more frequently than the tabu search during the searching processes, which implies that the chaotic search has higher ability than the tabu search as metaheuristics.
This paper proposes a dispersed generation system of multiple DC/DC converters with DC power sources connected in a ring formulation. Here is presented the analysis of the system based on the stored energy and passivity characteristics. Passivity Based Control (PBC), with its energy-modifying and damping-injection technique, is applied to a ring coupled converter system to stabilize itself at a desired DC voltage in the presence of external disturbances. The numerical results reveal the effective application of the control as a robust and flexible technique.