Correlated Information Extracting Learning using Neural Network

Summary

1. Humans can obtain useful information from sensor inputs and motion signals. The author takes a stand on the importance of the information which commonly exists among multiple inputs, which is called correlated information here, especially among sensor and motion signals. First, a method to get correlated information is proposed. In this method we enter multiple inputs to plural neural-networks respectively and make each output of each network communicate with that of the other network. Basic experiments were examined and it was confirmed that the correlated information can be extracted.

2. Spatial recognition ability can be obtained by learning from visual experiments. The author thinks that this ability is to extract correlated information between vision and motion signals. Learning to extract correlated information among multiple inputs can be done without supervisor. The author built a robot with a vision sensor in a computer and it was confirmed that the robot was able to recognize the distance vector of two dimensions from itself to a target object using neural-network without any supervisors.

Reference

1. Katsunari Shibata & Yoichi Okabe:
Unsupervised Learning of the Spatial Recognition Ability Using the Correlated Information Extracting Neural Network,
J. of JNNS(Japan Neural Network Society), Vol. 3, No. 1, pp.11-16, 1996. 5 (in Japanese)
柴田克成, 岡部洋一:

"相関情報抽出ネットと空間認識能力の教師なし学習"

日本神経回路学会誌, Vol. 3, No. 1, pp.11-16, 1996. 5

2. K. Shibata and Y. Okabe (柴田克成, 岡部洋一) :

"Learning of Multiple Dimensions of Information Extraction by Correlated Information Extracting Neural Network"
"相関情報抽出ニューラルネットによる複数次元抽出の学習"

ROBOMEC'97 講演論文集, Vol. B, pp.783-784, 1997.6

3. K. Shibata and Y. Okabe : PS File (2pages, 65KB, in Japanese)

"Unsupervised Learning to Extract the Depth of an Object on a Stereo Vision Using the Correlated Information Extracting Neural Network"
"相関情報抽出ネットによるステレオ画像上物体の奥行き情報抽出の教師なし学習"

4. K. Shibata : PS File (4pages, 50kB)

"A Neural-Network to get Correlated Information among Multiple Inputs"

Proc. of IJCNN'93 Nagoya, Vol.3, pp.2532-2535 (1993.10)

5. K. Shibata : PS File (4pages, 46kB)

"Spatial Recognition Model by Extracting Correlated Information

between Vision and Motion Information using Neural-Network"

Proc. of IJCNN'93 Nagoya, Vol.3, pp.2536-2539 (1993.10)