Neural Networks
A neural network is made of a number of processing elements also known as neuron whose inter-connection is synapsis (it is the link between two different neurons). Each neuron accepts the information from the external world or from the output of other neurons. The output signals from the neuron eventually propagate their effects across the entire network. The functionality and power of network primarily depend upon the number of neurons, inter-connectivity pattern and the value of weight assigned to each synapse. There are different types of neural networks and they are single layer feed forward neural network, multi-layer feed forward neural network and recurrent neural network. Single layer feed forward neural network consists of two layers and they are input layer and output layer. The input layer neurons receives the input signal and output layer neurons receives the output signal. The synaptic link carrying the weight connect the every neural to output neuron but not vise-verse. Single layer feed forward neural network is cyclic in nature. Despite the two layer the network is termed as single layer because the output layer alone performs all computation. The multi-layer feed forward neural network as the name implies is made up of multiple layers. Therefore the architecture of this neural network can have one or more intermediate layers which are also called hidden layers. The computational unit of intermediate layers is also known as hidden neurons which aids in performing useful information before directing the input to the output layers. The input layer neurons are linked to hidden layer neurons and the weight on these links are referred to as input-hidden layer weight. Similarly the hidden layer neurons are linked to the output layer neurons and the corresponding weight are referred to as hidden output layer weight. The recurrent neural network differs from the feed forward network architecture in the sense that there is at least one feedback loop.
Summary
A neural network is made of a number of processing elements also known as neuron whose inter-connection is synapsis (it is the link between two different neurons). Each neuron accepts the information from the external world or from the output of other neurons. The output signals from the neuron eventually propagate their effects across the entire network. The functionality and power of network primarily depend upon the number of neurons, inter-connectivity pattern and the value of weight assigned to each synapse. There are different types of neural networks and they are single layer feed forward neural network, multi-layer feed forward neural network and recurrent neural network. Single layer feed forward neural network consists of two layers and they are input layer and output layer. The input layer neurons receives the input signal and output layer neurons receives the output signal. The synaptic link carrying the weight connect the every neural to output neuron but not vise-verse. Single layer feed forward neural network is cyclic in nature. Despite the two layer the network is termed as single layer because the output layer alone performs all computation. The multi-layer feed forward neural network as the name implies is made up of multiple layers. Therefore the architecture of this neural network can have one or more intermediate layers which are also called hidden layers. The computational unit of intermediate layers is also known as hidden neurons which aids in performing useful information before directing the input to the output layers. The input layer neurons are linked to hidden layer neurons and the weight on these links are referred to as input-hidden layer weight. Similarly the hidden layer neurons are linked to the output layer neurons and the corresponding weight are referred to as hidden output layer weight. The recurrent neural network differs from the feed forward network architecture in the sense that there is at least one feedback loop.
Things to Remember
- A neural network is made of a number of processing elements also known as neuron whose inter-connection is synapsis (it is the link between two different neurons).
- Each neuron accept the information from the external world or from the output of other neurons. The output signals from the neuron eventually propagate their effects across the entire network.
- The functionality and power of network primarily depend upon the number of neurons, inter-connectivity pattern and the value of weight assigned to each synapse.
- There are different types of neural networks and they are single layer feed forward neural network, multi-layer feed forward neural network and recurrent neural network.
- Single layer feeds forward neural network consists of two layers and they are input layer and output layer. The input layer neurons receives the input signal and output layer neurons receives the output signal.
- Single layer feed forward neural network is cyclic in nature. Despite the two layer the network is termed as single layer because the output layer alone performs all computation.
- The multi layer feed forward neural network as the name implies is made up of multiple layers. Therefore the architecture of this neural network can have one or more intermediate layers which are also called hidden layers.
- The recurrent neural network differs from the feed forward network architecture in the sense that there is at least one feedback loop.
MCQs
No MCQs found.
Subjective Questions
Q1:
What is your nationality?
Type: Very_short Difficulty: Easy
Q2:
Define nation.
Type: Very_short Difficulty: Easy
Q3:
How is nation defined in the Interim Constitution of Nepal, 2063 BS?
Type: Short Difficulty: Easy
Q4:
What is national language according to our constitution?
Type: Very_short Difficulty: Easy
Q5:
What is nationality?
Type: Very_short Difficulty: Easy
Q6:
What should be done to protect our nationlity? Make a list.
Type: Short Difficulty: Easy
<ul>
<li>We must preserve our culture, tradition, and national properties.</li>
<li>All citizens should have a strong feeling of nationality.</li>
<li> All citizens should have unity among them.</li>
<li>Citizens should work for the development activities of the nation,</li>
</ul>
Q7:
Describe the relation between nation and nationality.
Type: Long Difficulty: Easy
Videos
No videos found.
Neural Networks
Neural Networks
It is considered to be the simplest model of biological normal system and therefore they have drawn their motivation from the kind of computing that is performed by human. The neural network in general is highly interconnected network of large number of processing element called neurons. The neural network can be massively parallel and exhibit characteristics such as mapping capabilities, fault tolerance, high speed information processing etc.
In simple terms a neural network is made of number of processing elements also known as neuron whose inter-connection is synapsis (it is the link between two different neuron). Each neuron accept the information from the external world or from the output of other neurons. The output signals from the neuron eventually propagate their effects across the entire network. The functionality and power of network primarily depends upon the number of neurons, interconnectivity pattern and the value of weight assigned to each synapses. There are different types of neural networks and they are as follows:
- Single layer feed forward neural network.
- Multi-layer feeds forward neural network.
- Recurrent neural network.
Single layer feed forward NN:
This type of neural network consists of two layers and they are input layer and an output layer. The input layer neurons receive the input signal and output layer neurons receives the output signal. The synaptic link carrying the weight connect the every neural to output neuron but not vise-versa.
Single layer feeds forward neural network is cyclic in nature. Despite the two layer, the network is termed as the single layer because the output layer alone performs all computation.
Multilayer feed forward NN:
The network as the name implies is made up of multiple layers. Therefore the architecture of this neural network can have one or more intermediate layers which are also called hidden layers. The computational unit of intermediate layers is also known as hidden neurons which aids in performing useful information before directing the input to the output layers. The input layer neurons are linked to hidden layer neurons and the weight on these links are referred to as input-hidden layer weight. Similarly the hidden layer neurons are linked to the output layer neurons and the corresponding weight are referred to as hidden output layer weight.
Recurrent NN:
This network differs from the feed forward network architecture in the sense that there is at least one feedback loop.
Comparison between artificial neural network (ANN) and biological neural network (BNN):
Parameters:
Speed:
| ANN | BNN |
| Neural networks are faster in processing information. The cycle time corresponding to the execution of one step of a program is in the order of Nano speed. | Biological neurons are slower in information processing. The cycle time corresponding to neural event promoted by an external stimulus occurs in the range of millisecond. |
Fault tolerance:
| ANN | BNN |
| Artificial neurons are inherently not fault tolerant because the information which is corrupted in the memory cannot be retrieved. | They exhibit fault tolerance because the information is distributed in the connection through the network. Even though if few connections are not working the information is still preserved due to distributed nature. |
Size and Complexity:
| ANN | BNN |
| They do not involve the computational unit very much. Therefore it is difficult to perform the complex pattern recognition. | They have large number of computing elements. Therefore the size and complexity of the connection gives the brain the power of performing complex pattern recognition task. |
Control Mechanism:
| ANN | BNN |
| There is a control unit which monitors all the activities of computing. | There is no central control for processing the information in the brain. There is no specific control mechanism external to the computing task. |
References:
- Elaine Rich, Kevin Knight 1991, "Artificial Intelligence".
- Nilsson, Nils J. Principles of Artificial Intelligence, Narosa Publishing House New Delhi, 1998.
- Norvig, Peter & Russel, Stuart Artificial Intelligence: A modern Approach, Prentice Hall, NJ, 1995
- Patterson, Dan W. Introduction to Artificial Intelligence and Expert Systems, Prentice Hall of India Private Limited New Delhi, 1998.
Lesson
Machine Learning
Subject
Computer Engineering
Grade
Engineering
Recent Notes
No recent notes.
Related Notes
No related notes.