JP2000122991A5 - - Google Patents
Download PDFInfo
- Publication number
- JP2000122991A5 JP2000122991A5 JP1998293690A JP29369098A JP2000122991A5 JP 2000122991 A5 JP2000122991 A5 JP 2000122991A5 JP 1998293690 A JP1998293690 A JP 1998293690A JP 29369098 A JP29369098 A JP 29369098A JP 2000122991 A5 JP2000122991 A5 JP 2000122991A5
- Authority
- JP
- Japan
- Prior art keywords
- layer
- neuron model
- vector
- point
- unit
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
- 210000002569 neurons Anatomy 0.000 description 11
- 230000000875 corresponding Effects 0.000 description 4
- 230000002860 competitive Effects 0.000 description 1
- 230000001537 neural Effects 0.000 description 1
Description
この態様では動作信号の入力に対する前記第2層ニューロンモデル群の各部分反応度が、まず算出される。そして、その部分反応度と、前記制御用ニューロンモデルと第2層ニューロンモデル群との間の結合荷重と、に基づいて、動作信号に対応する操作量を算出する。例えば前記制御用ニューロンモデルと第2層ニューロンモデル群との間の結合荷重を、前記部分反応度により重み付け加算して、操作量を算出すればよい。こうすれば、ニューラルネットワークの学習内容に基づき、動作信号に対応する操作量を算出することができる。 In this aspect, each partial reactivity of the second layer neuron model group with respect to the input of the motion signal is first calculated. Then, the operation amount corresponding to the operation signal is calculated based on the partial reactivity and the connection load between the control neuron model and the second layer neuron model group. For example, the operation amount may be calculated by weighting and adding the connection load between the control neuron model and the second layer neuron model group according to the partial reactivity. In this way, the amount of operation corresponding to the operation signal can be calculated based on the learning content of the neural network.
入力層10に学習ベクトルI1が与えられると、勝者ユニット(勝ちニューロン)、即ち学習ベクトルI1に対して最も反応する競合層12のニューロンモデルが決定される(S105)。具体的には、次式(3)により、競合層12のi番目のニューロンモデルと入力層10のニューロンモデルとの間のユークリッド距離D iを求め、それが最も短いものを勝者ユニットとする。 When the learning vector I 1 is given to the input layer 10, the winning unit (winning neuron), that is, the neuron model of the competing layer 12 that responds most to the learning vector I 1 is determined (S105). Specifically, the following equation (3), determine the Euclidean distance D i between the i-th neuron model and neuron model in the input layer 10 of the competitive layer 12, it is the winner units shortest.
図3には、上から学習ベクトル空間(平面)、重みベクトル空間(平面)、競合層12が順に描かれている。学習ベクトル空間は第1ベクトルX及び第2ベクトルYに対応する座標軸を有しており、学習ベクトルIが与えられると、それを平面上の一点に対応づけることができるようになっている。一方、重みベクトル空間は重みベクトルWの2成分に対応する軸を有しており、重みベクトルWiの終点が平面上の一点にそれぞれ描かれるようになっている。 In FIG. 3, the learning vector space (plane), the weight vector space (plane), and the competition layer 12 are drawn in this order from the top. The learning vector space has coordinate axes corresponding to the first vector X and the second vector Y, and when the learning vector I is given, it can be associated with a point on the plane. On the other hand, the weight vector space has an axis corresponding to the two components of the weight vector W, the end point of the weight vector W i is adapted to be drawn respectively to a point on the plane.
同図に示すように、ある学習ベクトルIが与えられると、競合層12に配置されるニューロンモデルから勝者ユニットが決定される。この勝者ユニットは、学習ベクトルIを重みベクトル空間上の一点に表した場合に、その点に最も近い点に終点を有する重みベクトルWiである。続いて、その勝者ユニットに対する近傍ユニットも選ばれる。同図では競合層12にニューロンモデルを平面的に配置しているため、勝者ユニットの周囲を取り巻くユニットが近傍ユニットに選ばれる。 As shown in the figure, given a certain learning vector I, the winning unit is determined from the neuron model arranged in the competing layer 12. The winner unit, when representing the learning vector I at one point on the weight vector space, a weight vector W i having an end point at a point closest to the point. Subsequently, a nearby unit with respect to the winning unit is also selected. In the figure, since the neuron model is arranged in a plane on the competing layer 12, the unit surrounding the winning unit is selected as the neighboring unit.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
JP29369098A JP4267726B2 (en) | 1998-10-15 | 1998-10-15 | Device for determining relationship between operation signal and operation amount in control device, control device, data generation device, input / output characteristic determination device, and correlation evaluation device |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
JP29369098A JP4267726B2 (en) | 1998-10-15 | 1998-10-15 | Device for determining relationship between operation signal and operation amount in control device, control device, data generation device, input / output characteristic determination device, and correlation evaluation device |
Publications (3)
Publication Number | Publication Date |
---|---|
JP2000122991A JP2000122991A (en) | 2000-04-28 |
JP2000122991A5 true JP2000122991A5 (en) | 2005-12-08 |
JP4267726B2 JP4267726B2 (en) | 2009-05-27 |
Family
ID=17797982
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
JP29369098A Expired - Fee Related JP4267726B2 (en) | 1998-10-15 | 1998-10-15 | Device for determining relationship between operation signal and operation amount in control device, control device, data generation device, input / output characteristic determination device, and correlation evaluation device |
Country Status (1)
Country | Link |
---|---|
JP (1) | JP4267726B2 (en) |
Families Citing this family (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP5034041B2 (en) * | 2006-07-20 | 2012-09-26 | 国立大学法人九州工業大学 | Data generation circuit and data generation method |
JP5130516B2 (en) * | 2006-09-05 | 2013-01-30 | 国立大学法人九州工業大学 | Data processing apparatus and method |
JP5011529B2 (en) * | 2006-11-01 | 2012-08-29 | 国立大学法人九州工業大学 | Data processing apparatus, data processing method, and program |
-
1998
- 1998-10-15 JP JP29369098A patent/JP4267726B2/en not_active Expired - Fee Related
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN100374175C (en) | Image generating device and method | |
CN107886167A (en) | Neural network computing device and method | |
CN107080940A (en) | Body feeling interaction conversion method and device based on depth camera Kinect | |
CN107247844A (en) | The minimum tree algorithms of X architecture Steiner based on adaptive PSO and mixing switching strategy | |
CN107357563B (en) | object moving method and device and storage medium, electronic device | |
CN110414138A (en) | The model of servo motor determines method and device, machinery equipment | |
US20240123348A1 (en) | Game character control method and apparatus, storage medium and electronic device | |
CN109701271A (en) | Show the methods, devices and systems of icon | |
Kose et al. | Q-learning based market-driven multi-agent collaboration in robot soccer | |
JP2000122991A5 (en) | ||
Salem et al. | Evolving a TORCS modular fuzzy driver using genetic algorithms | |
US6425826B1 (en) | Game machine and information storage medium | |
CN108837508A (en) | The method and apparatus of carrier are controlled in game | |
JPH07191951A (en) | Neural network circuit | |
Bell et al. | Ghost direction detection and other innovations for Ms. Pac-Man | |
KAKAZU | Learning coordinated motions in a competition for food between ant colonies | |
Sasamura et al. | A simple learning algorithm for growing ring SOM and its application to TSP | |
Zhou et al. | Cooperative behavior acquisition based modular Q learning in multi-agent system | |
Shi et al. | The most probable optimal design method for global optimization | |
Liebowitz et al. | Developing a neural network approach for intelligent scheduling in GUESS | |
JPH0342767A (en) | Learning system for neurocircuit network | |
Matsushiro | Considerations on the weighted minimum description length criterion | |
Roncancio et al. | Modeling virtual agent behavior in a computer game to be used in a real enviroment | |
CN115920400A (en) | Method and device for controlling non-player character, electronic device and storage medium | |
Mišeje et al. | Signal Processing in Smart Sensor Systems |