Tuesday, October 12, 2010

Neural Network

Neural Network:
Introduction:
            Whenever we talk about a neural network, we should more popularly say “Artificial Neural Network (ANN)”, ANN are computers whose architecture is modelled after the brain. They typically consist of hundreds of simple processing units which are wired together in a complex communication network. Each unit or node is a simplified model of real neuron which sends off a new signal or fires if it receives a sufficiently strong Input signal from the other nodes to which it is connected.
            Traditionally neural network was used to refer as network or circuit of biological neurones, but modern usage of the term often refers to ANN.
ANN is mathematical model or computational model, an information processing paradigm i.e. inspired by the way biological nervous system, such as brain information system. ANN is made up of interconnecting artificial neurones which are programmed like to mimic the properties of m biological neurons. These neurons working in unison to solve specific problems.
            ANN is configured for solving artificial intelligence problems without creating a model of real biological system. ANN is used for speech recognition, image analysis, adaptive control etc. These application are done through a learning process, like learning in biological system, which involves the adjustment between neurones through synaptic connection. Same happen in the ANN.


Working:
 Neural networks take a different approach to problem solving than that of conventional computers. Conventional computers use an algorithmic approach i.e. the computer follows the set of instructions in order to solve a problem. While neural networks process the information in a similar way the human brain does. Networks are composed of neurones which are working in parallel to solve a specific problem. Networks are learnt by example not programmed to perform specific task. The example must be selected carefully otherwise useful time is wasted or even worse the network might be functioning incorrectly. Thus, Its operation can be unpredictable.
Thus Neural networks & conventional algorithm computers not in competition but complement to each other. Means there are some tasks, more suited to computers while there are some tasks, which are more suited to the neural networks. In order to perform with maximum efficiency & large number of tasks. Conventional computer is used to supervise the neural networks.
Neurones:
                  1.      Simple neurones:
In this there are two modes of operations of neurones-
a)      Training Modes: The neuron can be trained to fire (or not), for particular Input patterns.
b)      Using Modes: When a taught input  pattern is detected at the input, its associated output becomes the current output.

2.      A more complicated neurones:
      It is a McCulloch and Pitts model (MCP). In these neurones, inputs are ‘weighted’; the effect that each input has at decision making is dependent on the weight of particular input. The weight of an input is a number which when multiplied with the input gives the weighted input. These weighted inputs are then added together and if they exceed a pre-set threshold value, the neuron fires. In any other case the neuron does not fire. In mathematical terms, the neuron fires if and only if;
X1W1 + X2W2 + X3W3 + ... > T
      The addition of input weights and of the threshold makes this neuron a very flexible and powerful one. The MCP neuron has the ability to adapt to a particular situation by changing its weights and/or threshold.

Architecture of neural networks:
a)      Feed-forward Network:
                                                Allow signals to travel from one way only, from Input to Output.

a)      Feedback Network:
                  In this signal travels in both the diarection by introducing loops in the network. These are powerful, dynamic & can get extremely complicated. their 'state' is changing continuously until they reach an equilibrium point.


Network layers
                       The commonest type of artificial neural network consists of three groups, or layers, of units: a layer of "input" units is connected to a layer of "hidden" units, which is connected to a layer of "output" units.

          The activity of the input units represents the raw information that is fed into the network.The activity of each hidden unit is determined by the activities of the input units and the weights on the connections between the input and the hidden units.The behaviour of the output units depends on the activity of the hidden units and the weights between the hidden and output units.
Transfer Function
The behaviour of an ANN (Artificial Neural Network) depends on both the weights and the input-output function (transfer function) that is specified for the units. This function typically falls into one of three categories:


·         linear (or ramp)
·         threshold
·         sigmoid
For linear units, the output activity is proportional to the total weighted output.
For threshold units, the output is set at one of two levels, depending on whether the total input is greater than or less than some threshold value.
For sigmoid units, the output varies continuously but not linearly as the input changes. Sigmoid units bear a greater resemblance to real neurones than do linear or threshold units, but all three must be considered rough approximations.

Applications:

Real life applications
The tasks to which artificial neural networks are applied tend to fall within the following broad categories:
·         Function approximation, or regression analysis, including time series prediction and modelling.
·         Classification, including pattern and sequence recognition, novelty detection and sequential decision making.
·         Data processing, including filtering, clustering, blind signal separation and compression.
·         Application areas of ANNs include system identification and control (vehicle control, process control), game-playing and decision making (backgammon, chess, racing), pattern recognition (radar systems, face identification, object recognition, etc.), sequence recognition (gesture, speech, handwritten text recognition), medical diagnosis, financial applications, data mining (or knowledge discovery in databases, "KDD").


Advantages:
1.      Adaptive learning: An ability to learn how to do tasks based on the data given for training or initial experience.
2.      Self-Organisation: An ANN can create its own organisation or representation of the information it receives during learning time.
3.      Real Time Operation: ANN computations may be carried out in parallel, and special hardware devices are being designed and manufactured which take advantage of this capability.





(  
This article is the topic of unit "IT Trends" from Nagpur University syllabus MBA 3 rd Sem IT  notes.

Topics of this unit are is covered in other blog post. Links are given below.
Topics Covered

  1. Biometrics
  2. GIS-Geographical Information System
  3. Google Earth
  4. Audio Visuals MPEG
  5. I-Pod
For more notes you can also refer to other links as given below 
 
)

2 comments: