|
Neural Network Fundamentals is a two-day course that presents the basic concepts of neural computing and its implementation in MATLAB to the participants. Fundamental topics of
neural networks are introduced, ranging from a brief introduction to the history of neural
computing, concept of a single neuron, introduction to supervised neural networks such as
perceptrons, linear networks, backpropagation networks and radial basis networks, to the
introduction of unsupervised neural network, i.e. the self-organizing map. Hands-on
demonstration and exercises are vital element of the course, with heavy emphasis on practical
applications of neural networks. The ultimate aim of this course is to instill full appreciation
of the powerful capability of MATLAB and the Neural Network Toolbox for the
implementation of neural computing.
Pre-requisites:
The pre-requisites for this course are MATLAB Fundamentals and Programming Techniques,
and experience with basic computer operations. The MATLAB for Signal Processing course
is strongly recommended, and having a basic knowledge of neural network related concepts
would be a plus.
Outline:
Introduction
Objective: Understand The MathWorks Products, brief company history, and course schedule.
• Course outline
• Course materials
Course Outline
Day 1
Topic 1: Introduction to MATLAB
Introduction to basic features and user-interfaces of MATLAB. Understand The MathWorks
Products and brief company history
i) The MathWorks Product Family
ii) The MATLAB Desktop
iii) Help features in MATLAB
Topic 2: Neural Network: A Brief Introduction
Presents a brief history of neural network and its MATLAB terminology:
i) What is Neural Network
------a. Neural Network Application
------b. Definition of Neural Network
------c. Biological prospective of Neural Networks
ii) Simple Neuron Model
------a. MATLAB representation of the simple neuron model.
iii) Architecture of neural network
iv) Data structures
Topic 3: Perceptrons
Learn about the simplest form of neural network and its applications:
i) Introduction
------a. Linearly separable problems
------b. The perceptron neuron
------c. MATLAB representation of the perceptron neuron.
ii) The perceptron architecture
iii) Creating a perceptron
------a. The Common-line approach.
iv) Perceptron learning rule
v) Training of perceptron
vi) Examples
vii) Exercises
Topic 4: Linear Networks
Learn about a more advance form of neural network for solving linearly-separable problems:
i) Introduction
------a. Linear Neuron
------b. MATLAB representation of the perception Neuran.
ii) The architecture of Linear Networks
iii) The Widrow-Hoff learning Algorithm
iv) Linear clarification
v) Adaptive filtering
------a. Designing adaptive filtering
vi) Examples
vii) Exercise
Day 2
Topic 5: Backpropagation Networks
Learn about one of the most popular neural networks, renowned for its strength in solving
highly complex and non-linear problems:
i) Introduction
ii) Architecture of Feedforward BP network
------a. MATLAB representative of Feedforward BP network
------b. Transfer function of BP networks
iii) Learning algorithms for backpropagation networks
------a. Training of backpropagation network
------b. Batch gradient descent training
------------i. Batch gradient descent with momentum
------c. Faster training
------d. Comparison of Training algorithm
------------i. Improving generalization with early stopping
iv) Preprocessing and postpossessing
v) Examples
vi) Exercise
Topic 6: Self-Organizing Maps
Learn about the most popular unsupervised neural network for feature extraction and data
mining:
i) Introduction to self-organizing maps
ii) Competitive Learning
------a. Learning algorithm for competitive learning
iii) Self-organizing maps
------a. Topologies and Distance Function
------b. SOM architecture
iv) Training of self-organizing maps
------a. Ordering phase
------b. Tuning phase
v) Examples
vi) Exercise
Topic 7: Radial Basis Networks
Learn about an alternative form of neural network to backpropagation networks:
i) Introduction
ii) Radial basis Neuron Model
iii) Generalized Regression Networks
iv) Probabilistic Neural Networks
v) Examples
vi) Exercise
Trainer Profile
Please kindly check with our Training Consultants for more
details.
|
