@article{oai:bunkyo.repo.nii.ac.jp:00003416,
 author = {鈴木, 昇一},
 journal = {情報研究, Information and Communication Studies},
 month = {1993-01-01, 2012-01-17},
 note = {When a correspondence between no r m alized prices which appear on Walras' law in the microeconomics and similarity-measures in a field of pattern-information processing is taken into account, a process of obtaining from an old simslarity-measure a new similarity-measure which is better in a faculty of separation and clustering than the old one may be explained by the assumption that a quantity which corresponds to an excess demand function is expressed in terms of a Hopfield neural network. In the present paper, a mechanism of accommodating the similarity-measure function to a set of training patterns is described and is related to a supervised learning process with a steepest descent strategy.
\n　価格調整過程を記述する方程式の(均衡価格)について成り立つワルラスの法則と同様なことが、パターン情報処理における類似度の調整過程において成立するならば、類似度関係からseparation・clusteringにつき性能の良い今一つの類似度関数が得られ、しかも、各財への超過需要量に対応する量をHopfield neural netの形式で表現すると、類似度関数の教師あり学習過程も記述され得ることが示されている。},
 pages = {211--236},
 title = {ミクロ経済学におけるワルラスの法則とパターン類似度関数のホップフィールドニューラルネット形調整(情報学共同研究)},
 volume = {14},
 year = {}
}