Title:
INTER-NEURON COUPLING STRUCTURE ESTIMATING METHOD
Document Type and Number:
Japanese Patent JP3179196
Kind Code:
B2
Abstract:
PURPOSE: To accurately estimate the inter-neuron coupling structure by comparing conditional probabilities to distinguish the correlations between positive and negative to quantitatively detect the magnitude of correlations between time series data.
CONSTITUTION: Data indicating the times when active potentials of a neuron X on the transmission side and a neuron Y on the reception side are generated are inputted to generate a correlation histogram of the relative time difference between both of these times. The correlation histogram is so standardized that the sum of probabilities is a proportion of generation of the active potential of the neuron Y for generation of the active potential of the neuron X, thereby calculating the probability in which the neuron Y generates the active potential with a time difference (t) at the time of generation of the active potential of the neuron X, namely, a conditional probability p(Y1,t&verbar X1). A conditional probability p(X1,t&verbar X0) is calculated by the probability of generation of the active potential of the neuron Y, and other conditional probabilities are calculated. If the conditional probability p(Y1,t&verbar X0) is smaller than p(Y1,t&verbar X0 the negative sign is added to a communication capacity C; but otherwise, the positive sign is added to it.
Inventors:
Satoshi Yamada
Kenji Matsumoto
Satoru Shiono
Kenji Matsumoto
Satoru Shiono
Application Number:
JP20017092A
Publication Date:
June 25, 2001
Filing Date:
July 06, 1992
Export Citation:
Assignee:
Mitsubishi Electric Corporation
International Classes:
G06F15/18; G06G7/60; G06N3/02; G06N3/04; G06N99/00; (IPC1-7): G06N3/04; G06G7/60
Other References:
【文献】1990年電子情報通信学会春季全国大会講演論文集 1990 7.225 山田訓 他「ニューラルネットを用いる光学他店同時測定データからの活動電位検出」
Attorney, Agent or Firm:
Hiroaki Tazawa (2 outside)