In this article, we will take you through a complete journey in Artificial Neural Networks, their complex versatilities and the benefits of using them in our daily lives to build Artificial Intelligence through Machine Learning.
An Artificial Neural Network (ANN) is a computational algorithmic model. It is based on the structure of biological neural networks within human brains and aim to replicate their functions in order to augment human activities. It ideally aims to work in the way human brains processes information. It includes a large number of connected processing units that work together to process information. They also generate meaningful results from it.
“The neural network is this kind of technology that is not an algorithm, it is a network that has weights on it, and you can adjust the weights so that it learns. You teach it through trials.” - Howard Rheingold (AI Author)
Structural Breakdown of Artificial Neural Networks (ANNs)
As mentioned above, Artificial Neural Networks or ANNs are intended to simulate the behavior of biological systems composed of “neurons”. ANNs are computational models inspired by an animal’s central nervous systems. It is capable of machine learning as well as pattern recognition. These are presented as systems of interconnected “neurons” which can compute values from inputs.
A neural network is basically an oriented graph. It consists of nodes which in the biological analogy represent neurons, connected by arcs. It corresponds to dendrites and synapses. Each arc associated with a weight while at each node. Apply the values received as input by the node and define Activation function along the incoming arcs, adjusted by the weights of the arcs, as shown in the diagram below.
In the above figure we can see that an Artificial Neural Network is a Machine Learning algorithm based on the model of a human neuron. The human brain consists of millions of neurons. It sends and process signals in the form of electrical and chemical signals. These neurons are connected with a special structure known as synapses. Synapses allow neurons to pass signals. From large numbers of simulated neural networks. This is how the human brain processes information. Each neuron is dedicated to fulfilling a set of tasks based on a specific electrochemical threshold which needs to be achieved in order to pass on the information to the next level.
Similarly, as shown in the diagram, an Artificial Neural Network is modelled on the same principles where each input layer is connected to a successive series of output layers based on mathematically assigned weights and biases which act as thresholds and thus resemble a similar information processing technique. It works like the way human brains process information and this enables the network to reach a desired set of goals based on set decision-making parameters at every successive stage.
Furthermore, the networks possess a unique built-in capability to alter or modify these goals at each successive stage with the sole purpose of maximizing success for a given set of tasks; like the human brain. This constitutes the fundamental basis of Machine Learning in the domain of Artificial Intelligence. ANNs include a large number of connected processing units that work together to process information. They also generate meaningful results from it. So we can apply Artificial Neural Networks not only for classification, but also for regression of continuous target attributes.
An Artificial Neural Network must always contain the following 3 layers:
Input layer – The activity of the input units represents the raw information that can feed into the network.
Hidden layer – To determine the activity of each hidden unit. The activities of the input units and the weights on the connections between the input and the hidden units. There may be one or more hidden layers.
Output layer – The behavior of the output units depends on the activity of the hidden units and the weights between the hidden and output units.
Types of Artificial Neural Networks (ANNs)
There are several kinds of Artificial Neural Networks. These types of networks are primarily classified based on different mathematical operations and difference in the set of parameters required to determine the output. Let’s look at some types of Neural Networks that are widely used today:
1. Feedforward Neural Network: This neural network is one of the simplest forms of ANNs known today, where the data or the input travels only in one direction.
2. Radial basis function Neural Network: Radial basic functions consider the distance of a point with respect to the centre. RBF functions have two layers, first where the features are combined with the Radial Basis Function in the inner layer and then the output of these features are taken into consideration while computing the same output in the next time-step which is basically a memory.
3. Kohonen Self Organizing Neural Network: The objective of a Kohonen map is to input vectors of arbitrary dimension to discrete map comprised of neurons. The map needs to me trained to create its own organization of the training data. It comprises of either one or two dimensions.
4. Modular Neural Network: Modular Neural Networks have a collection of different networks working independently and contributing towards the output. Each neural network has a set of inputs which are unique compared to other networks constructing and performing sub-tasks. These networks do not interact or signal each other in accomplishing the tasks. The advantage of a modular neural network is that it breakdowns a large computational process into smaller components decreasing the complexity.
5. Recurrent Neural Network (RNN): Recurrent Neural Networks work on the principle of Long-Short Term Memory or LSTMs thereby saving the output of a layer and feeding this back to the input to help in predicting the outcome of the layer.
6. Convolutional Neural Network (CNN): Convolutional neural networks are similar to feed forward neural networks, where the neurons have learnable weights and biases. Its application has been primarily in signal and image processing which takes over OpenCV in field of computer vision.
Advantages & Disadvantages of Neural Networks
Let’s us now discuss a few advantages and disadvantages of using ANNs:
1. Neural networks perform well with linear and nonlinear data but a common criticism of neural networks, particularly in robotics, is that they require a large diversity of training for real-world operation. This is so because any learning machine needs sufficient representative examples in order to capture the underlying structure that allows it to generalize to new cases.
2. Neural networks work even if one or few units fail to respond to network but to implement large and effective software neural networks, much processing and storage resources need to be committed. While the brain has hardware tailored to the task of processing signals through a graph of neurons, simulating even a most simplified form on Von Neumann technology may compel a neural network designer to fill millions of database rows for its connections – which can consume vast amounts of computer memory and hard disk space.
3. Neural networks learn from the analyzed data and do not require reprogramming but they are referred to as black box” models, and provide very little insight into what these models really do. The user just needs to feed it input and watch it train and await the output with zero control over the "hidden-layers".
Neural Networks can find great applications in data mining once integrated with Big-Data Analytics Platforms & SaaS products which and can be used extensively in sectors such as Social Media, Space & Astronomy, High-Energy Particle Physics, Stock-market trading, Economics, Government Welfare, Internet of Things etc. Moreover, its pattern recognition capabilities can also revolutionize the fields of Health-care, Astronomy, Life Sciences and Education across the world today as ANNs can easily be used for insight driven Big-data Analytics and data classification problems in large datasets and data-streams after careful training and optimization.
Thus, on the whole, ANNs are considered as simple but promising mathematical models that are efficient enough to enhance existing Big-Data Analysis technologies. Although it is not comparable with the power of the human brain today, it definitely serves as the basic building blocks for Artificial Intelligence which holds the promise of surpassing human intelligence within our very lifetime!
Stay tuned for my next blog post explaining each type of Neural Network in greater depth along with their exciting technological applications in our modern-day lives!!
Comments