Introduction • Class description • Class policy • Goals & topics of the course • McCulloch-Pitts neurons

History of ANNs • Selected ANN applications • Artificial neural networks – brief history • McCulloch-Pitts neurons, examples

Moscow campus closed due to sever inclement weather. Classes were canceled.

Single neuron operation • Selecting rectangular area • Network architecture • Matlab code • The effect of gain on control surface (soft activation function) • (if there is time) Selecting any two-dimensional area • Net/Out values by neurons & layers and gain effect • Partitioning in 3-dim space • Review - Analytic geometry in Euclidean space with Cartesian coordinates • slope intercept, line intercept, scalar form of a line, 2 & 3 point form, one point and vector form • Calculating distance in Euclidean space • Linear machine and minimum distance classifiers • Bisecting approach

Calculating distance in Euclidean space • Linear machine and minimum distance classifiers • Bisecting approach • Winner Take All • Here the class was interrupted due to power outage in Idaho Falls

HW#1&2 review • XOR problem, two neurons • xy ; x+y ; | xy ; |( xy ); |( x+y ) • NNOTT • Neural Network Online Training Tool • Gradient Descent Training • Steepest Descent Method • Newton Method • Levenberg -Marquardt Algorithm (LM Algorithm)

Feedback networks • Associative Memories • Hopfield Autoassociative Memories • Bi-directional Associative Memories (BAM) • Multidirectional Associative Memories • Dynamic Feedback Networks • Implementation • AD converter • Optimization problems (traveling salesman) • Finding function minimum • State variable approach

Counter Propagation Networks (CPN) • Analog memories • RBF – Radial Basis Function Networks • More on Competitive Networks • LVQ – Learning Vector Quantization • Matching & Self-Organizing Networks • MAXNET • Hamming Network

RBF – Radial Basis Function Networks • More on Competitive Networks • LVQ – Learning Vector Quantization • Matching & Self-Organizing Networks • MAXNET • Hamming Network • Cascade Correlation networks

Fuzzy Logic • Introduction, membership degree • AND, OR, NOT • Entropy • Subsethood • Composition • Min-Max; Product-Sum; Product-Max ; • Fuzzy Associative Memories (FAM) • correlation product (max of products) • correlation minimum encoding (max of min)
Supplementary material (Session 26 from Intelligent Control Course, Fall 06): Properties of Fuzzy Sets • alpha-cut, support, core, height, level • Fuzzy Numbers – Special Case of Fuzzy Sets • Definition and properties • Fuzzy Arithmetic • Arithmetic operations on intervals Crisp interval analysis

Fuzzy controller • Zadeh min-max controller • Fuzzification • Fuzzy inference engine (rule table) • Defuzzification • Zadeh fuzzy controller • design example, inputs, outputs (singletons)
Supplementary material (Session 24 from Intelligent Control Course, Fall 06): Crisp (hard) clustering • k -Means clustering • Fuzzy clustering • c -Means clustering, steps • Constraints & cluster count • Fuzzy adaptive clustering • Main principle, steps • Normalization and hedge dilution • Generating Rule Prototypes • Clusters into rules

Tagagi-Sugeno fuzzy controller • design example, inputs, outputs • Fuzzy Systems • VLSI Implementation • Microprocessor Implementation • Error Comparison of various fuzzy controllers • Neural Systems • Elliott's Activation Function • Neuro Controller • Microprocessor Implementation • Comparison of Fuzzy & Neuro Controllers • Pulse Coded Neural Networks

VLSI Implementation of ANNs • Multilevel logic multiplier • example: 4x4, 6x6 • Refinement of Fuzzy Operators • classification – general & functional • behavior – illustration • Fuzzy Decision Support Systems (FDSS) • fuzzy preference relation • Genetic Algorithms, Evolutionary Computation • general terms & concepts • initialization, selection, reproduction (with crossover & mutation) • examples • GANNs • Final exam preparation

Final exam preparation • Class paper presentations

Class paper presentations

Class paper presentations