Chapter4_Part3 DifferentialHeb learning Differential Competitive learning Tutor:Prof.Gao Reporter:WangYing
Chapter4_Part3 Differential Heb learning & Differential Competitive learning Tutor : Prof. Gao Reporter : WangYing
Review 一P >Signal Heb Learning Law my =-my +S,S, Competitive Learning Law riny =S,[S,-mg ring =-S my +S,S, 2006.10.30
2006.10.30 Review ➢ Signal Heb Learning Law ➢ Competitive Learning Law m m S S ij ij i j = − + m S m S S ij j ij i j = − + m S S m ij j i ij = −
Part I:Differential Heb Learning Learning law m,=-m,+S,S,+S,S Its simpler version m,=-m,+S,S Hebbian correlations promote spurious causal associations among concurrently active units. Differential correlations estimate the concurrent and presumably causal variation among active units. 2006.10.30
2006.10.30 Part I: Differential Heb Learning Learning law Its simpler version Hebbian correlations promote spurious causal associations among concurrently active units. Differential correlations estimate the concurrent and presumably causal variation among active units. m m S S S S ij ij i j i j = − + + m m S S ij ij i j = − +
Differential Heb Learning >Fuzzy Cognitive Maps (FCMs >Adaptive Causal Inference >Klopfs Drive Reinforcement Model >Concomitant Variation as Statistical Covariance >Pulse-Coded DifferentialHebbian Learning 2006.10.30
2006.10.30 Differential Heb Learning ➢ Fuzzy Cognitive Maps (FCMs) ➢ Adaptive Causal Inference ➢ Klopf’s Drive Reinforcement Model ➢ Concomitant Variation as Statistical Covariance ➢ Pulse-Coded Differential Hebbian Learning
Fuz2 y Cognitive Maps(模糊认知映射 Fuzzy signed directed graphs with feedback.It model the world as a collection of classes and causal relations between classes. C:Se瓜of computers C:Profits The directededge efrom causal concept C,to concept C measures fow much C causes C 2006.10.30
2006.10.30 Fuzzy Cognitive Maps (模糊认知映射) Fuzzy signed directed graphs with feedback. It model the world as a collection of classes and causal relations between classes. The directed edge from causal concept to concept measures how much causes . ij e Ci Cj Ci Cj Ci Cj ij e : Sells of computers : Profits Ci Cj