ARTIFICIAL NEURAL NETWORK

Mar 19 • General • 3787 Views • 5 Comments on ARTIFICIAL NEURAL NETWORK

ARTIFICIAL NEURAL NETWORK

An Artificial neural network (ANN) is an artificial representation of the human brain that tries to simulate its learning process. It is simply called a ’Neural Net (NN) ’which is interconnected group of artificial neurons that uses a mathematical model or computational model for information processing based on a connectionist approach to computation. ANN is a network of simple processing elements (artificial neurons) which exhibit complex global behavior, determined by the connections between the processing elements and element parameters that changes its structure based on external or internal information flows through the network.

The conventional computers are not good for –interacting with noisy data or data from environment, fault tolerance and where we cannot formulate an algorithmic solution. NN are a form of multiprocessor system, with:

1) Simple processing elements,
2) A high degree of interconnection
3) Simple scalar messages and
4) Adaptive interaction between elements. So ANN can take decisions for non-algorithmic problems and noisy data.

Working Principle:

It works on threshold logic i.e. all the input are summed up and if the input is greater than threshold logic then output is 1 otherwise is 0.

Output= sgn (∑input i-Ω)

if ( ∑input i>Ω) then output=1 ;

if (∑ input<Ω) then output=0

Architecture of ANN

where the inputs are given a certain weight .

ANN architecture:

Single Layer Feed-forward Network: It consists of a single layer of weights, where the inputs are directly connected to the output via a series of weights.

Multi Layer Feed-forward Network: This type of architecture consists of intermediate layer besides of input layer and output layer known as HIDDEN LAYER.

Use of Hidden Layers:

1) It does intermediate computation before directing the input to the output layers
2) The input layer and output layer neurons are linked with hidden layer neurons.
3) Output layer neuron=l-m-n

where , l=input neurons;
m=neurons in hidden layers
n=output neurons

Recurrent Network: In this system there is a feedback path loops i.e. the output of a neuron is feedback into itself as input.

Learning Methods of ANN:

On the basis of presence and absence of teacher and the information provided for the system to learn the learning process is divided into three parts:

a) Supervised Learning: This learning process is based on comparison between current output and expected output and then generated error is corrected.  A teacher is present with expected output and the network is trained with every input pattern.
b) Unsupervised Learning: Neither teacher nor expected output pattern is given. NN learns from its own experiences.
c) Reinforced Learning: Teacher is present but does not present the expected output, only indicates that the output is right or wrong.

There are some other learning processes of ANN:

(a)    Back Propagation Method (b) Genetic Algorithm Method  (c) Competitive Method etc

Application of ANN:

• Learning the distribution of Object Trajectories for Event Recognisition: The techniques being developed will allow models of characteristic object behaviors to be learnt from the continuous observation of long image sequences. It is hoped that these models of characteristic behaviors will have a number of uses, particularly in automated surveillance and event recognition, allowing the surveillance problem to be approached from a lower level, without the need for high-level scene/behavioral knowledge. The model is learnt in an unsupervised manner by tracking objects over long images sequences,and is based on a combination of a neural network implementing Vector Quantization and a type of neuron with short-term memory capabilities.The model is learnt in an unsupervised manner by tracking objects over long image sequences,and is based on a combination of a neural network implementing . Vector Quantization and a type of neuron with short-term memory capabilities .

• Radiosity for Virtual Reality Systems (ROVER): The synthesis of actual and computer generated photo-realistic images has been the aim of artists and graphic designers for many decades. Some of the most realistic images were generated using radiosity techniques.

ROVER MODELLING : Autonomous Walker & Swimming Eel: The research in this area involves combining biology, mechanical engineering and information technology in order to develop the techniques necessary to build a dynamically stable legged vehicle controlled by a neural network.This would incorporate command signals, sensory feedback and reflex circuitry in order to produce the desired movement. Simulation is done by ANN.

• Robocup: Robot World Cup: The RoboCup Competition pits robots (real and virtual) against each other in a simulated soccer tournament. The aim of the RoboCup competition is to foster an interdisciplinary approach to robotics and agent-based AI by presenting a domain that requires large-scale corporation and coordination in a dynamic, noisy, complex environment.
• Artificial Life: Galapagos : Galapagos is a fantastic and dangerous place where up and down have no meaning, where rivers of iridescent acid and high-energy laser mines are beautiful but deadly artifacts of some other time. Through spatially twisted puzzles and bewildering cyber-landscapes, the artificial creature called Mendel struggles to survive, and you must help him.
  Mendel is a synthetic organism that can sense infrared radiation and tactile stimulus. His mind is an advanced adaptive controller featuring Non-stationary Entropic Reduction Mapping—a new form of artificial life technology developed by Anark. He can learn like dog, he can adapt to hostile environment like a cockroach, but he cannot solve the puzzles that prevent his escape from Galapagos..
• Detection and Tracking of Moving Targets: (Defense Group Incorporated) : The moving target detection and track methods here are “track  before detect” methods. They correlate sensor data versus time and location, based on the nature of actual tracks.The track statistics are “learned” based on artificial neural network (ANN) training with prior real or simulated data. Effects of different clutter backgrounds are partially compensated based on space-time-adaptive processing of the sensor inputs, and further compensated based on the ANN training. Specific processing structures are adapted to the target track statistics and sensor characteristics of interest. Fusion of data over multiple wavelength and sensors are also supported.The system localizes and tracks peoples’ faces as they move through a scene. It integrates the  following scene:
  1. Motion Detection
  2. Tracking people based upon motion
  3. Tracking faces using an appearance model
  • Behavioral Animation and Evolution of Behavior: This is a classic experiment and the flocking of “boids,” that convincingly bridged the gap between artificial life and computer animation.Each boid has direct access to the whole scene’s geometric description, but reacts only to flock mates within a certain small radius of itself. The basic flocking model consists of following behaviors:
  1.  Separation: steer to avoid crowding local flock mates.
  2. Alignment: steer towards the average heading of local flock mates.
  3. Cohesion: steer to move toward the average position of local flock mates.

  In addition, the more elaborate behavioral model included predictive obstacle avoidance and goal seeking. Obstacle avoidance allowed the boids to fly through simulated environments while dodging static objects. For applications in computer animation, a low priority goal seeking behavior caused the flock to follow a scripted path.

• Past and Present  :  The development of true Neural Networks is a fairly recent event, which has been met with success. Two of the different systems (among the many) that have been developed are: the basic feed forward Network and the Hopfield Net.In addition to the applications featured here, other application areas include:

1. Financial Analysis — stock predictions.
2. Signature Analysis — the banks in America have taken to NNs to compare signatures with what is stored.
3. Process Control Oversight — NNs are used to advise aircraft pilots of engine problems
4. Direct Marketing — NNs can monitor results from a test mailing and determine the most successful areas.

• The Future : The future of Neural Networks is wide open, and may lead to many answers and/or questions. Is it possible to create a conscious machine?

« Sample Paper for NIFT Entrance Test with Solutions Sample paper of microprocessor »