-
[show abstract]
[hide abstract]
ABSTRACT: In this article, we propose a midpoint-validation algorithm for a support vector machine which improves the generalization
of the support vector machine so that the midpoint-validation error is minimized. We compared its performance with other techniques
for support vector machines, and also tested our proposed method on fifth benchmark problems. The results obtained from the
simulation shows the effectiveness of the proposed method.
Key wordsSupport vector machine-Midpoint-validation-Pattern classification problem
Artificial Life and Robotics 04/2012; 15(2):138-141.
-
[show abstract]
[hide abstract]
ABSTRACT: The support vector machine (SVM) is known as one of the most influential and powerful tools for solving classification and
regression problems, but the original SVM does not have an online learning technique. Therefore, many researchers have introduced
online learning techniques to the SVM. In this article, we propose an unsupervised online learning method using a self-organized
map for a SVM. Furthermore, the proposed method has a technique for the reconstruction of a SVM. We compare its performance
with the original SVM, the supervised learning method for the SVM, and a neural network, and also test our proposed method
on surface electromyogram recognition problems.
Artificial Life and Robotics 04/2012; 14(3):362-366.
-
Neural Information Processing - 18th International Conference, ICONIP 2011, Shanghai, China, November 13-17, 2011, Proceedings, Part II; 01/2011
-
[show abstract]
[hide abstract]
ABSTRACT: The goal of Human-Computer Interface (or called Human-Robot interface) research is to provide humans with a new communication
channel that allows translating people’s intention states via a computer into performing specific actions. This paper presents
a novel hands-free control system for controlling the electric wheelchair, which is based on Bio-signals as surface electromyogram
signals. The Bioelectric signals are picked up from facial muscles then the Bio-signals are passed through an amplifier and
a high pass filter. Motion control commands (Forward, Left, Right, Forward to the Right, Forward to the left and Stop) are
classified by simple rule. These commands are used for controlling the electric wheelchair.
KeywordsSurface Electromyogram Signal-Electric Wheelchair-Motion Control Command Rule-The Obstacle Evading Function
10/2010: pages 97-104;
-
[show abstract]
[hide abstract]
ABSTRACT: In this paper, an improved guided genetic algorithm is proposed for the job-shop scheduling problem. The proposed method is improved by a genetic algorithm using multipliers which can be adjusted during the search process. Simulation results based on some benchmark problems demonstrate that the proposed method can find better solutions than the genetic algorithm and the original guided genetic algorithm. © 2010 Wiley Periodicals, Inc. Electron Comm Jpn, 93(8): 15–15, 2010; Published online in Wiley InterScience (www.interscience.wiley.com). DOI 10.1002/ecj.10263
Electronics and Communications in Japan 07/2010; 93(8):15 - 22. · 0.08 Impact Factor
-
IEICE Transactions. 01/2010; 93-D:2073-2079.
-
IEICE Transactions. 01/2010; 93-A:936-941.
-
[show abstract]
[hide abstract]
ABSTRACT: Research surface electromyogram (s-EMG) signal recognition using neural networks is a method which identifies the relation
between s-EMG patterns. However, it is not sufficiently satisfying for the user because s-EMG signals change according to
muscle wasting or to changes in the electrode position, etc. A support vector machine (SVM) is one of the most powerful tools
for solving classification problems, but it does not have an online learning technique. In this article, we propose an online
learning method using SVM with a pairwise coupling technique for s-EMG recognition. We compared its performance with the original
SVM and a neural network. Simulation results showed that our proposed method is better than the original SVM.
Artificial Life and Robotics 02/2009; 13(2):483-487.
-
ISMVL 2009, 39th International Symposium on Multiple-Valued Logic, 21-23 May 2009, Naha, Okinawaw, Japan; 01/2009
-
IEICE Transactions. 01/2008; 91-A:3385-3388.
-
[show abstract]
[hide abstract]
ABSTRACT: In this paper, a model of the neuron based on the dendrite mechanism is proposed. The neuron model takes the form of a dendrite, and the interaction between synapses and so on, which are not considered by the conventional model, is taken into account. In this paper, the form of a dendrite is assumed to be arbitrary and interaction is assumed to exist among all synapses. That is, a superfluous synapse is formed, and is expressed in a logical network, then the equivalence between them is shown. The ripened dendrite is formed during learning by screening unnecessary dendrites, and strengthening and fixing the useful dendrites. The model proposed in this paper can correspond to the linearly unseparated problem, unlike the one-layer traditional neuron. Moreover, learning of this model suggests the form of the dendrites and the positions of the excitation and inhibition synapses. This may give some very useful suggestions for biologists. © 2001 Scripta Technica, Electron Comm Jpn Pt 3, 84(8): 11–24, 2001
Electronics and Communications in Japan (Part III Fundamental Electronic Science) 04/2001; 84(8):11 - 24. · 0.14 Impact Factor
-
[show abstract]
[hide abstract]
ABSTRACT: In this paper, we propose a hill-climbing learning method for Hopfield networks in which the energy of the network is intentionally raised in the weight space so that the network can escape from local minima. This learning method involves repeated updating of the Hopfield network in state space and modification of the weights in weight space if it settles into a local minimum, so that the energy is raised. Shifting of the global minimum during learning can be avoided by updating the state on the unlearned Hopfield network in state space, using equilibrium states obtained from learning as the initial state. Simulations using this learning method show that the Hopfield network can escape from a local minimum which depends on the initial state. A simulation experiment is first conducted with a two-variable Hopfield network, and the convergence of learning and the change of the domain from a local minimum to a global minimum are demonstrated. Then, using an example of the traveling-salesman problem, a simulation is performed using concrete numerical values, and the validity of the method is demonstrated. © 2001 Scripta Technica, Electron Comm Jpn Pt 3, 84(7): 28–40, 2001
Electronics and Communications in Japan (Part III Fundamental Electronic Science) 03/2001; 84(7):28 - 40. · 0.14 Impact Factor
-
[show abstract]
[hide abstract]
ABSTRACT: this paper we haveattempted to improve the performance of current-mod MVL circuits using #MOS transistors. First a low-voltageand lowpower #MOS current mirror is presented Second a current-mod #MOSthreshold dreshol and a T-gate are proposed as applications of MVL circuits. The improved
09/1999;
-
[show abstract]
[hide abstract]
ABSTRACT: A current-mode CMOS analog defuzzification circuit which has already proposed is insensitive to absolute variation of the fabrication process and it is easy to expand into a multiple-input one. However, it has two disadvantages. One is that it needs two input currents which are equal in magnitude and opposite in direction (sign). A current steering circuit which consists of current mirrors is used to change the direction of the input current signals. However, because current mirrors are strongly influenced by the channel length modulation effect, it is very difficult to generate the desired current signal. The other disadvantage is that it needs input bias currents for all input signals, which causes high power consumption. In this paper, a current-mode CMOS analog four-quadrant multiplier for defuzzification circuits is proposed. The proposed multiplier is insensitive to absolute variation of the fabrication process. Furthermore, it removes the above disadvantages. A defuzzification circuit using the proposed multiplier is also designed. The proposed defuzzification circuit has decreased power dissipation owing to the use of a common bias block circuit. These circuits are analyzed in detail with the PSPICE program. Lastly, the proposed defuzzification circuit is evaluated and the validity is discussed. © 1997 Scripta Technica, Inc. Electron Comm Jp Pt 3, 80(4): 30–41. 1997
Electronics and Communications in Japan (Part III Fundamental Electronic Science) 12/1998; 80(4):30 - 41. · 0.14 Impact Factor
-
01/1995
-
[show abstract]
[hide abstract]
ABSTRACT: The local linear wavelet neural network is an improvement of wavelet net-work and commonly used learning algorithm is gradient descent method. In this paper, we attempt to predict sunspots, Mackey-Glass time series and Box-Jenkins data using a local linear wavelet neural network. Furthermore, we propose a technique using bi-directional function learning method. The simulation results show the effectiveness of the proposed method.
-
[show abstract]
[hide abstract]
ABSTRACT: In this paper, we propose a midpoint-validation method, which improves the generalization of neural networks. The problem associated with the former cross validation method is that efficiency is affected due to the separation of training data into two or more set. As for the proposed method, it creates midpoint data from the known training data and calculates a set of criteria using the newly created midpoint data and the previous training data. The implementation is easy since there is no unnecessary processing involved in separating the data into two or more sets. The advantage of the proposed method is that the method becomes much more efficient compared to the former method due to the numerical simulation used. We compare its performance with those of the Support Vector Machine (abbr. SVM), Multilayer Perceptron (abbr. MLP), Radial Basis Function (abbr. RBF) and the proposed method was tested on several benchmark problems. The results obtained from the simulation carried out shows the effectiveness of the proposed method.
Innovative Computing ,Information and Control, International Conference on.
-
寛夫 熊谷,
寛幸 谷川,
公一 淡野,
興彦 石塚,
Hiroo Kumagai,
Hiroyuki Tanigawa, Koichi Tanno,
Okihiko Ishizuka,
ヒロオ クマガイ,
ヒロユキ タニガワ,
コウイチ タンノ,
オキヒコ イシズカ
[show abstract]
[hide abstract]
ABSTRACT: Abstract In recent years, the world of wireless communications has been changing very rapidly. The wireless communication devices need RF analog circuits such as mixer. In CMOS technology, if it successfully merges RF analog and baseband digital blocks, it is possible to reduce size, weight,cost, and power consumption. However, the circuit elements which can be operated under low supply voltage are required because of lowering of the breakdown voltage with the development of fabrication technique in CMOS process. In this paper, a low-supply-voltage and wide-bandwidth CMOS analog multiplier is proposed, which is indispensable for mixer. The proposed multiplier has advantages of low voltage operation and low noise. Simulations of the multiplier demonstrate the input range of 0.54 Vp-p, the -3 dB bandwidth of 1.03 GHz and the S/N ratio of 131 dB with VDD=2 V. Next, the proposed multiplier is applied to a downconversion mixer. The mixer can be operated at VDD=2V, and the conversion gain and 3rd order intcrcept point (IIP3) are -O.8 dR and +10.9 dBm, respectively.
-
幸博 生島,
昭一郎 石川,
公一 淡野,
興彦 石塚,
Yukihiro Ikushima,
Shouichirou Ishikawa, Koichi Tanno,
Okihiko Ishizuka,
ユキヒロ イクシマ,
ショウイチロウ イシカワ,
コウイチ タンノ,
オキヒコ イシズカ
[show abstract]
[hide abstract]
ABSTRACT: Power electronic devices are widely used in industries and commercial fields. Recently high-speed M0S devices have been exchanged instead of SCR. In this paper, we design a bias circuit of a MOS switch with high accuracy. The circuit for stabilizing on-resistance of the MOS FET at input voltage 3 to 5 [V] and temperature -50 to 85 [℃] is designed. The bias circuit consists of a CR oscillator, a charge pump circuit with the 4-time voltage and a constant voltage circuit. A CR oscillator contains resistors, capacitors and a CMOS inverter. The oscillation frequency is determined by the values of C and R. By using the oscillation frequency as a control clock, a charge pump circuit with the 4-time voltage is operated as expected. Using the 4-time voltage of the charge pump circuit, the constant voltage circuit holds the output voltage of 9.9 [V] without influence of the input voltage and a room temperature. Finally, the designed MOS switch has a low and accurate on-resistance. Their performances are verified using the HSPICE simulator with 0.5μm CMOS process device parameters.
-
[show abstract]
[hide abstract]
ABSTRACT: Abstract A multiple-valued logic(MVL) circuit is a basic circuit which performs MVL digital processing. In this paper, a MVL latch circuit are presented, that holds a given MVL level. Originally, the MVL latch has been proposed by Mirmotahari, et. al. We improve it's characteristics with neuron MOSFETs. The noise margin is 30.0[m.Ⅴ] compared with 48.0[mⅤ] in the original circuits.
