Neural networks

Neural network is columniation of algorithms that helps to recognize the relationships in set of data through a special process that working as human brain operators. So it indicates to system of neurons ,it can be organic or artificial in nature.

Neural networks from the base of deep learning. It sub filled machine learning where the algorithms are inspired the structure of human brain. Neural networks taking data. Train themselves recognize the patterns in this data and predict the outputs for in new set of similar data. Neural networks are made of layers of neurons. These neurons are the co-processing units of the network.

IDENTIFICATION

First we have the input layer which receives the output. So the output layers predict our final output.

    In between there is hidden layers which performs most of the computations required by out network.             

HOW IT WORKS

Here is the image of circle. It compose 28* 28 pixels = 784 pixels. Each pixels pare as inputs to each neurons of first layer. Neurons of one layer is connect to next layer. Through channels. Each of channels has numerical value known as weight. The inputs are multiply by co responding weights and the sum is sent as input to the neurons in the hidden layer. Each of these neurons is associate with the numerical value call the buyers which is add to the input sum.

So this value is pass through the threshold function call the activation function. Result of the activation function determines if the particular neuron will get activate or not. Activated neurons transmits data to the neurons of the particular neurons of the next layer over the channels. This manner the data is propagate through the network. This is call forward propagation . In the output layer the neurons with highest value fires in determines the outputs. The values fires in determines the outputs. The values are basically aid probability.

Our neural network is associate with squires has the highest value of probability. Hence that the output predicted by the neural network. Of cause just look at it, we know the neural network has made a wrong prediction.

BUT HOW DOES THE NEURAL NETWORK FIGER THIS OUT?

Note that our network is yet to be train during the process along with the input our network. Also as the output fed to it. The predict out put is compare against the actual output to realize the error prediction. The magnitude to the error indicates how wrong we are. And the sign suggested our predicted value are the higher or lower than we expected. The errors here giving indication of direction and magnitude of change to reduce the error. This information is then transport backward thought our network known as back propagating. Based on this information the weights are adjust. This cycle of forward propagation and Bach propagation is perform with multiple inputs. This process continuous till our weight are assign the network can predict the shapes correctly. In most of cases.

TIME CONSUMING

Neural networks may take hours or even months to train.But time is a reasonable trade off when compared to its scope.

PRIMAL APPLICATIONS OF NEURAL NETWORKS

• facial recognition
• Forecasting
• Music composition

These processes are still in primary stage.Those google, Amazon,Nvidia companies invested to develop such As libraries,predictive models  trough neural networks.

TYPES OF NEURAL NETWORKS

• Artificial neural network
• recurrent neural network
• convolutional neural network
• feedforward neural network
• Multilayer perception
• perceptron
• Redial basis function network
• Modular neural network
• Long  short term memory
• Autoencoder

Neural network is columniation of algorithms that helps to recognize the relationships in set of data through a special process that working as human brain operators. So it indicates to system of neurons ,it can be organic or artificial in nature. Neural networks from the base of deep learning. It sub filled machine learning where…