## Introduction to Probabilistic Neural Network

The probabilistic neural network could be a feedforward neural network; it is widely employed in classification and pattern recognition issues. PNN has three layers of nodes. In the PNN algorithmic program, the parent likelihood distribution performance of every category is approximated by a Parzen window and a non-parametric perform.

### Layers of Probabilistic Neural Network

The operations of the PNN square measure organized into a multilayered feedforward network with totally different layers. There square measure four layers. They are:

- Input layer
- Pattern layer
- Summation layer
- Output layer

**Input Layer: **We predict a value, and it is given to the input layer where it subtracts the median. These input neurons are fed with the values to the next layer, which is the hidden layer to each hidden node.

Predictor variable terribly very cell is delineated by the input layer. Then the input neurons feed the values to each of the neurons at intervals the hidden layer.

**Pattern Layer: **This layer contains one cell in each case at intervals the coaching job data values. It stores the value of the predictor variables for the case on the aspect of the target worth. A hidden cell computes the geometrician distance of the action at law from the neurons centre purpose then applies the radial basis perform kernel perform victimization the letter of the alphabet values.

**Summation Layer: **In the target variable, there is one pattern for each category of PNN. The actual target category every coaching job case is held on with each hidden cell. The weighted value commencing of a hidden cell is fed alone to the pattern cell that corresponds to the hidden neurons category. The pattern neurons add the values for the class they represent.

**Output Layer: **The output layer compares the weighted votes for each target accumulated at intervals the pattern layer and uses the foremost necessary vote to predict the target category.

4.5 (4,812 ratings)

View Course

### Applications of Probabilistic Neural Network

- It is employed in modelling structural deterioration of stormwater pipes.
- It is employed in finding totally different pattern classification issues.
- It helps in population pharmacology.
- It helps in remote sensing image classification.
- Ship identification victimizationPNN.

### Algorithm for PNN

The following steps set up the PNN:

- Sorting them into N variety of sets wherever every set contains one category of vectors.
- For every set, we have a tendency to outline the mathematician perform.
- For every set, we have a tendency to outline the summed mathematician output perform.

Once the PNN is outlined, then we are able to feed vectors into it and classify them as follows:

- Read the input vector and feed it to every mathematician to perform in each category.
- For every cluster of hidden nodes, reckon all mathematician useful values at the hidden nodes.
- For every cluster of hidden nodes, feed all its mathematician useful values to the one output node for that cluster.
- At every category output node, total all of the inputs and multiply by constant.
- Find the most worth of all summed useful values at the output nodes.

We principally use PNN rather than multilayer perceptron, by victimization the PNN we’ve got a few blessings and downsides. We’ve got mentioned some.

### The architecture of Probabilistic Neural Network

Below is the architecture of PNN:

In the above diagram, it represents the architecture of the PNN. A probabilistic neural network has three nodes. The above diagram represents the 2 class PNN, but it can be extended to X number of classes. The input layer has N number of nodes. These N number of nodes are connected with a branch to the hidden layer so that the hidden node received the complete input feature.

The hidden nodes are collected into groups for each X class. Each hidden node in the group for class X corresponds to a Gaussian function centred on its associated feature in the class. These Gaussian functional values are sent to the output layer, and they are called the X output nodes.

### Advantages & Disadvantages of PNN

Below are the advantages and disadvantages of Probabilistic Neural Network.

**Advantages:**

- PNNs square measures a lot of quicker than multilayer perceptron networks.
- PNNs may be a lot of correct than multilayer perceptron networks.
- PNN networks square measure comparatively insensitive to outliers.
- PNN networks generate correct foreseen target likelihood.
- PNNs approach Bayesoptimum.

**Disadvantages:**

- PNN is slower as compared to a multi-layer perceptron.
- PNN needs a lot of memory areas to store.

### Conclusion

We have discussed the probabilistic neural network. we also discussed the Layers of PNN, Applications of PNN, Algorithm of PNN, Architecture of PNN, Advantages on PNN, and Disadvantages of PNN.

### Recommended Articles

This is a guide to Probabilistic Neural Network. Here we discuss the Introduction and its architecture of Probabilistic Neural Network along with Advantages as well as disadvantages. You can also go through our suggested articles to learn more –