Scribd Logo
Hochladen

Welcome to Scribd!

  • AI010 804L01 Neural Networks

    Uploaded by

    varsha raichal
    36 views41 pages
    This document provides an overview of artificial neural networks (ANNs). It defines ANNs as simplified models of the brain that can perform functions like classification, prediction, and noise reduction. ANNs learn from examples through training and can generalize to new inputs. Common network architectures include single and multi-layer feedforward networks and recurrent networks. Learning methods include supervised, unsupervised, and reinforced learning. Examples of early ANN models are the perceptron and ADALINE/MADALINE networks. The document also discusses applications of ANNs in areas like pattern recognition, optimization, forecasting, and control systems.

    Original Description:

    Original Title

    mod1_Neural Networks

    Copyright

    © © All Rights Reserved

    Available Formats

    PPTX, PDF, TXT or read online from Scribd

    Did you find this document useful?

    Is this content inappropriate?

    Report this Document
    This document provides an overview of artificial neural networks (ANNs). It defines ANNs as simplified models of the brain that can perform functions like classification, prediction, and noise reduction. ANNs learn from examples through training and can generalize to new inputs. Common network architectures include single and multi-layer feedforward networks and recurrent networks. Learning methods include supervised, unsupervised, and reinforced learning. Examples of early ANN models are the perceptron and ADALINE/MADALINE networks. The document also discusses applications of ANNs in areas like pattern recognition, optimization, forecasting, and control systems.

    Copyright:

    © All Rights Reserved
    Download as PPTX, PDF, TXT or read online from Scribd
    Flag for inappropriate content
    Download as pptx, pdf, or txt
    36 views41 pages

    AI010 804L01 Neural Networks

    Uploaded by

    varsha raichal
    This document provides an overview of artificial neural networks (ANNs). It defines ANNs as simplified models of the brain that can perform functions like classification, prediction, and noise reduction. ANNs learn from examples through training and can generalize to new inputs. Common network architectures include single and multi-layer feedforward networks and recurrent networks. Learning methods include supervised, unsupervised, and reinforced learning. Examples of early ANN models are the perceptron and ADALINE/MADALINE networks. The document also discusses applications of ANNs in areas like pattern recognition, optimization, forecasting, and control systems.

    Copyright:

    © All Rights Reserved
    Download as PPTX, PDF, TXT or read online from Scribd
    Flag for inappropriate content
    Download as pptx, pdf, or txt
    of 41

    AI010 804L01 Neural Networks

    Module 1
    What Are Artificial Neural Networks?
     An extremely simplified model of the brain/biological
    neuron system
     Essentially a function approximator
     Transforms inputs into outputs to the best of its ability
     It is a massively parallel distributed processing system
    madeup of highly interconnected neural computing
    elements that have the ability to learn and there by acquire
    knowledge and make it available for use

     Various learning mechanisms are available to enable NN


    acquire knowledge
     Based on the learning mechanisms, NN architectures are
    classified into different types.
     Learning process – training
     The ability to solve problem using the knowledge acquired –
    inference

     NN are simplified imitation of the central nervous system and


    has been motivated by the kind of computing performed by
    human brain

     Hence the technology can be termed as


     Artificial Neuron Systems(ANS) technology
     Artificial Neural Networks(ANN)
     Neural Networks (NN)
    What Are They Used For?

     Classification
    Pattern recognition
    feature extraction
    image matching

    Noise Reduction
    Recognize patterns in the inputs and produce noiseless
    outputs

    Prediction
    Extrapolation based on historical data
    Why Use Neural Networks?
     Ability to learn
     NN’s figure out how to perform their function on their own
     Determine their function based only upon sample inputs

     Ability to generalize
     i.e. produce reasonable outputs for inputs it has not been taught
    how to deal with
    Human Brain
     Composed of many “neurons” that co-operate to perform the
    desired function
     It is the increased neural activity which is responsible for
    learning and memory

     Donald Hebb “ when an axon of a cell A is near enough to


    excite a cell B and repeatedly or persistantly takes part in
    firing it , some growth process or metabolic change takes
    place in one or both cells such that A’s efficiency as one of
    the cells firing B is increased”
    Model of Artificial Neuron
    Summation unit thresholding unit
    x1
    w1
    inputs
    x2
    weights

    I = w1x1+w2x2+w3x3+ .............+wnxn
    Activation functions
     To generate the final output Y , the sum is passed through a
    non linear fiter Ø called activation function or transfer
    function or squash function

    Y= Ø (I)
    a) Thresholding function
    sum is compared with a thershold value
    b) Signum function
    also called quantizer function
    c) Sigmoidal function
    Continuous function varies gradually between the
    asymptotic values 0 and 1 or -1 and +1
    Differentiable
    d) Hyperbolic tangent function
    it can produce negative output values
    Typical architectures
     Represented by directed graph
     Graph G is an ordered 2- duple (V,E) consisting of a set V
    of vertices and a set E of edges

    v1 v3
    e1

    e4 v5
    e2
    e5
    v2 v4
    e3
    Single layer feedforward network

    Muti layer feed forward network

    Recurrent network
    Characteristics of neural networks
    Mapping capabilities - they can map i/p pattern to
    associated o/p patterns
    NN learn by examples – NNs can be trained with known
    examples of a problem before they are tested for their
    inference capability on unknown instances of problem
    Capability to generalize – can predict new outcomes from
    past trends
    Robust systems and fault tolerant – they can recall full
    pattern from the incomplete , partial or noisy pattern
    Can process information in parallel at high speed and in a
    distributed manner.
    Training and learning methods
    Training methods
    Supervised
    Unsupervised
    Reinforced
    Learning Rules
    1. Hebbian learning
     Based on corrective weight adjustments
     W = ∑ Xi Yi T

    2. Gradient descent learning


     ∆Wij = ɳ dE/dWij
    Examples :
    widrow
    Hoffs delta rule
    Backpropagation learning rule

    3. Competitive learning
    “Winner – takes – all strategy”

    4. Stochastic learning
    Weights are adjusted in probabilistic manner
    Classification Learning algorithm

    Supervised unsupervised reinforced(o/p based)


    (error based)

    stochastic
    Error correction hebbian competitive
    (gradient descent)

    back propagation
    least mean square
    Taxonomy of NN architectures
     ADALINE ,ART
     AM ,BAM
     Boltzmann Machine ,BSB
     CCN,Cauchy Machine
     Hopfield Network, LVQ
     MADALINE, MLFF
     Neocogniton,Perceptron
     RBF, RNN, SOFM
    LEARNING METHOD
    TYPE GRADIENT HEBBIAN COMPETITIVE STOCHASTI
    OF DESCENT C
    ARCHIT SS FF ADALINE AM LVQ -
    ECTURE Hopfield Hopfield SOFM
    Percepron
    ML CCN Neocognito - -
    FF MLFF n
    RBF
    RN RNN BAM ART Boltzmann
    BSB machine
    Hopfield Cauchy
    machine
    Early neural network architectures
    Rosenblatt’s perceptron
    Computational model of the retina of the eye.
    Sensory unit association unit response unit
    Basic learning algorithm for training
    perceptron
    1. Wij (k+1) = Wij k
    2. Wij (k+1) = Wij k -ɳxi
    3.
    Wij (k+1) = Wij k +ɳxi

    Limitation
    Can handle problems with linear separable solution
    space
    Linear sereparability
    X - OR problem

    The problem for ANN is to identify the input as odd parity and the even
    parity
    Why Perceptron is unable to find solution for
    nonlinear problem
    Perceprton learning rule

    Simplified Rule
    Theorem
    ADALINE Network
    MADALINE network
    Applications of ANNs
    1. Pattern Recongnition(PR)/ Image processing
    2. Optimisation / constraint satisfaction
    3. Forecasting & Risk management
    4. Control systems

    You might also like