http://chineseinput.net/에서 pinyin(병음)방식으로 중국어를 변환할 수 있습니다.
변환된 중국어를 복사하여 사용하시면 됩니다.
Complexity Control Method of Chaos Dynamics in Recurrent Neural Networks
Masao Sakai,Noriyasu Homma,Kenichi Abe 제어·로봇·시스템학회 2002 International Journal of Control, Automation, and Vol.4 No.2
This paper demonstrates that the largest Lyapunov exponent λ of recurrent neural networks can be controlled efficiently by a stochastic gradient method. An essential core of the proposed method is a novel stochastic approximate formulation of the Lyapunov exponent λ as a function of the network parameters such as connection weights and thresholds of neural activation functions. By a gradient method, a direct calculation to minimize a square error (λ-λ^obj)^2, where λ^obj is a desired exponent value, needs gradients collection through time which are given by a recursive calculation from past to present values. The collection is computationally expensive and caused unstable control for the exponent for networks with chaotic dynamics because of chaotic instability. The stochastic formulation derived in this paper gives us an approximation of the gradients collection in a fashion without the recursive calculation. This approximation can realize not only a faster calculation of the gradient, but also stable control for chaotic dynamics. Due to the non-recursive calculation, without respect to the time evolutions, the running times of this approximation grow only about is N^2 compared to as N^5 T that is of the direct calculation method. It is also shown by simulation studies that the approximation is a robust formulation for the network size and that proposed method can control the chaos dynamics in recurrent neural networks efficiently.
Complexity Control Method of Chaos Dynamics in Recurrent Neural Networks
Sakai, Masao,Homma, Noriyasu,Abe, Kenichi Institute of Control 2002 Transaction on control, automation and systems eng Vol.4 No.2
This paper demonstrates that the largest Lyapunov exponent λ of recurrent neural networks can be controlled efficiently by a stochastic gradient method. An essential core of the proposed method is a novel stochastic approximate formulation of the Lyapunov exponent λ as a function of the network parameters such as connection weights and thresholds of neural activation functions. By a gradient method, a direct calculation to minimize a square error (λ - λ$\^$obj/)$^2$, where λ$\^$obj/ is a desired exponent value, needs gradients collection through time which are given by a recursive calculation from past to present values. The collection is computationally expensive and causes unstable control of the exponent for networks with chaotic dynamics because of chaotic instability. The stochastic formulation derived in this paper gives us an approximation of the gradients collection in a fashion without the recursive calculation. This approximation can realize not only a faster calculation of the gradient, but also stable control for chaotic dynamics. Due to the non-recursive calculation. without respect to the time evolutions, the running times of this approximation grow only about as N$^2$ compared to as N$\^$5/T that is of the direct calculation method. It is also shown by simulation studies that the approximation is a robust formulation for the network size and that proposed method can control the chaos dynamics in recurrent neural networks efficiently.
KIM, JONG-SHIK,SAKAI, MASAO,LEE, SI-KYUNG,YAHANG, CHAHNG-SOOL,MATSUGUCHI, TATSUHIKI 한국미생물 · 생명공학회 2001 Journal of microbiology and biotechnology Vol.11 No.1
In order to investigate the role of bacterial chemotaxis in root colonization, the chemotaxis potential of bacteria isolated from spinach roots was compared with that of bacteria from nonrhizosphere soil, with reference to the plant age (1,000isolates), soil moisture conditions (1,400isolates), and part of the root (200isolates). The % CT (% occurrence of chemotaxis (+) isolates among total bacterial isolates) of the root isolates significantly fluctuated during the plant growth period, reaching a maximum after 10-15 days of growth. At this time period, the maximum % CT for the root isolates was around 70-80% CT under a soil moisture of 50% WFP (% volume of waterfilled pores in total soil pores), and then gradually reduced with an increasing % WFP. The results of the chemotaxis potential of each of the 100 isolates from the spinach roots and nonrhizosphere soil under various % WFP demonstrated that the % CT of the root isolates were significantly higher than those of isolates from the nonrhizosphere soil under a wide range of soil moisture content (35-80% WFP). Furthermore, the % CT value (80%) from the upper root was significantly higher than that (55%) from the lower root. Compared with the % CT values of the roots, the values from the nonrhizosphere soil did not significantly vary relative to the plant age or % WFR These results indicate that chemotaxis would appear to be a major factor in bacterial root colonization.
Auto-Detection of Non-Isolated Pulmonary Nodules Connected to The Chest Walls in X-ray CT images
Satoshi Shimoyama,Noriyasu Homma,Masao Sakai,Tadashi Ishibashi,Makoto Yoshizawa 제어로봇시스템학회 2009 제어로봇시스템학회 국제학술대회 논문집 Vol.2009 No.8
In this paper, we develop an auto-detection method of non-isolated pulmonary nodules for computer-aided diagnosis (CAD) of lung cancers using X-ray CT images. An essential core of the method is to transform the non-isolated nodules connected to the walls of the chest into isolated ones that can be detected more easily by CAD systems developed previously. To this end, an active contour model is proposed to extract the lung area from the original CT image. The proposed model can solve the local optimum problem of the contour model by using anatomical features of the lung in X-ray CT slices. Some experimental results demonstrate the usefulness of the proposed method by using clinical CT images.