# Different types of Neural Network with its Architecture

**Introduction:-**

Neural network are simplified model of the biological nervous system. Neural network is a highly interconnected network of a large number of processing elements called neurons in an architecture inspired by the brain. Neural networks exhibit characteristics such as mapping capabilities or pattern association, generalization, fault tolerance and parallel and high speed information processing. This network adopt various learning mechanism. This network learn by examples and thus architecture can be trained with known example of a problem.

**Benefits of Neural network:-**

In neural network several parallel processing operation is done and thus evaluate the example faster. The Neural Network is robust on noisy training data. It posses the capability to generalize, that is, they can predict new outcome from past trends. It also exhibit mapping capabilities, that is, they can map input pattern to their associated output patterns.

**Model of Neuron:-**

Here, weight models the acceleration of the input signals. Weights are the multiplicative factors of the inputs to account for the strength of the synapse. To generate the final output y, sum is passed on to a non linear filter called activation function or transfer function. Commonly used activation function is the thresholding function where sum is compared with a threshold value and if the value of total input is greater than threshold value, then output is 1 else it is 0.

**Neural network as directed graph:-**

Neural network structure can be represented using a directed graph. A graph is consisting of a set of vertices and set of edges. When each edge is assigned an orientation, the graph is called a directed graph. In neural network, the significance of graph is as signal are restricted to flow in specific directions. The vertices represent input/output and edges represent the links.

**Network Architecture:-**

Neural network architecture have been classified as:

**Single layer feed forward networks,****Multilayer feed forward networks and****Recurrent networks**

**Artificial Neural Network:-**

Neural networks are simplified imitation of central nervous system and it is motivated by the kind of computation which is also performed by human brains. The structural constituent of human brain is neuron, which perform computation. The technology which has been built on simplified imitation of computing by neurons of brain is called Artificial Neural Network.

**Learning Processes:**

**1. Hebbian learning-**

It is based on correlative weight adjustment and is proposed by Hebb. Here the input and output pattern are associated by the weight matrix.

**2.** **Competitive learning-**

Here, weights are updated for those neuron which respond strongly to the input stimuli. When input pattern is presented all neuron in the layer compete and the winning neuron undergoes weight adjustment.

**3.** **Error correction learning-**

It is based on the minimization of error defined in terms of weight and the activation function of the network. It is also required that the activation function is differentiable, as the weight update is dependent on the gradient of the error.

**4.** **Stochastic learning-**

Here, weights are adjusted in a probabilistic fashion. An example is simulated annealing-the learning mechanism employed by Boltzmann and Cauchy machines.

**Linearly separable task:-**

Set of point in 2D space are linearly separable if set can be separated by the straight line. In general, a set of points in n-dimensional space are linearly separable if there is a hyperplane of (n-1) dimension that separates the sets.

**Learning Process:**

In associative mapping, the network learns to produce a particular pattern on the set of input units whenever another particular pattern is applied on the set of input units.

In regularity detection, units learn to respond to particular properties of the input patterns.

Statistical learning theory, also goes by other names such as nonparametric classification and estimation, supervised learning and statistical pattern recognition and the Pattern recognition can be implemented by using a feed-forward neural network. During training, the network is trained to associate outputs with input patterns. When the network is used, it identifies the input pattern and tries to output the associated output pattern.

**Related Questions and Answers:**

**Q1. Define neural network.**

Ans.- Neural network is simplified central nervous system model and it is motivated by the computing which is performed by human brains and the technology built on this logic is termed as neural network.

**Q2. What are the benefits of neural network?**

Ans.- Benefits are as follows-

Parallel computing, faster, robust on noisy training data, mapping capabilities.

** Q3. What is directed graph?**

Ans.- We know graph consist of vertices and edges and when orientation is assigned to edge of graph then it is called as directed graph.

**Q4. Classify neural network architecture.**

Ans.- Neural network architecture is classified as – single layer feed forward networks, multilayer feed forward networks and recurrent networks.

**Readers can give their suggestions / feedbacks in the given below comment section to improve the article.**

**Related Topics:**

Introduction To Artificial Intelligence

Questions about Neural Networks

### Tell us Your Queries, Suggestions and Feedback

### 3 Responses to Different types of Neural Network with its Architecture

« Different types of Data Link Layer- DLL Protocols Perceptron with its Optimization Techniques »

mc,xzc

mc,xzc