ISSN: 2394-3122 (Online) e-ISJN: A4372-3088 Impact Factor: 3.471 Volume 4, Issue 4, April 2017
SK International Journal of Multidisciplinary Research Hub
Journal for all Subjects
Research Article / Survey Paper / Case Study Published By: SK Publisher (www.skpublisher.com)
Basic Application and Study of Artificial Neural Networks
Md. Tanjil Sarker1
Department of Computer Science and Engineering, Jagannath University,
Bangladesh
Sanjida Noor2 Department of CSE,
University Information Technology & Sciences, Bangladesh
Dr. Uzzal Kumar Acharjee3 Chairman & Associate Professor,
Dept. of CSE, Jagannath University
Bangladesh
Abstract: In this paper, we are expounding Artificial Neural Network or ANN, its different qualities and business applications. In this paper we additionally demonstrate that "what are neural systems" and "Why they are so essential in today's Artificial knowledge?" Because various advances have been made in creating Intelligent framework, some roused by natural neural systems. ANN gives an exceptionally energizing choices and other application which can assume imperative part in today's software, Computer engineering field. There are a few Limitations likewise which are said. An Artificial Neural Network (ANN) is a data handling worldview that is motivated by the way natural sensory systems, for example, the mind, prepare data. The key component of this worldview is the novel structure of the data preparing framework. It is made out of an extensive number of exceptionally interconnected handling components (neurons) working as one to take care of particular issues. ANNs, similar to individuals, learn by illustration. An ANN is designed for a particular application, for example, design acknowledgment or information arrangement, through a learning procedure. Learning in natural frameworks includes conformity to the synaptic associations that exist between the neurons. This is valid for ANNs too. This paper gives outline of Artificial Neural Network, working and preparing of ANN. It additionally clarifies the application and points of interest of ANN.
Key Words: ANN(Artificial Neural Network), Neurons, pattern recognition, Feedback Network, Feed Forward Network, Artificial Neuron, Characteristics and Application.
I. INTRODUCTION
The idea of ANN is essentially presented from the subject of science where neural system plays an imperative and key part in human body. In human body work is finished with the assistance of neural system. Neural Network is only a web of bury associated neurons which are millions in number. With the assistance of these interconnected neurons all the parallel handling is done in human body and the human body is the best case of Parallel Processing. A neuron is an exceptional natural cell that procedure data starting with one neuron then onto the next neuron with the assistance of some electrical and substance change.
It is made out of a cell body or soma and two sorts of out achieving tree like branches: the axon and the dendrites. The cell body has a core that contains data about genetic characteristics and plasma that holds the atomic hardware's or creating material required by the neurons. The entire procedure of accepting and sending signs is done specifically way like a neuron gets signals from other neuron through dendrites. The Neuron send signals at spikes of electrical movement through a long thin stand known as an axon and an axon parts this signs through neurotransmitter and send it to alternate neurons Neural systems, with their
Volume 4, Issue 4, April 2017 pg. 1-12 striking capacity to get importance from confused or uncertain information, can be utilized to concentrate designs and recognize patterns that are too perplexing to be in any way saw by either people or other computer methods.
Fig 1Structure of typical Neurons
A trained neural network can be thought of as an "expert" in the category of information it has been given to analyses.
Other advantages include:
1. Adaptive learning: A capacity to figure out how to do undertakings in view of the information given for preparing or starting knowledge.
2. Self-Organization: An ANN can make its own particular association or portrayal of the data it gets amid learning time.
3. Real Time Operation: ANN calculations might be completed in parallel, and exceptional equipment gadgets are being planned and fabricated which exploit this capacity
4. Fault Tolerance via Redundant Information Coding: Fractional decimation of a system prompts to the relating corruption of execution. Be that as it may, some system abilities might be held even with real system harm.
II. ARTIFICIAL NATURAL WORK
Neural network adopt an alternate strategy to critical thinking than that of conventionalcomputers. Traditional computers utilize an algorithmic approach i.e. the Computer takes after an arrangement of guidelines keeping in mind the end goal to take care of an issue. Unless the particular strides that the computer needs to take after are known the computer can't take care of the issue. That confines the critical thinking capacity of ordinary computers to issues that we as of now comprehend and know how to understand. However, computers would be quite a lot more helpful in the event that they could do things that we don't precisely know how to do. Neural systems handle data correspondingly the human cerebrum does. The system is made out of countless interconnected preparing components (neurons) working in parallel to tackle a particular issue. Neural systems lear n by illustration. They can't be customized to play out a particular assignment. The cases must be chosen deliberately generally valuable time is squandered or surprisingly more terrible the system may work inaccurately. The drawback is that on the grounds that the system discovers how to take care of the issue independent from anyone else, its operation can be eccentric.
Then again, ordinary computers utilize an intellectual way to deal with critical thinking; the way the issue is to explain mu st be known and expressed in little unambiguous guidelines. These directions are then changed over to an abnormal state dialect program and afterward into machine code that the computer can get it. These machines are absolutely unsurprising; in the event that anything turns out badly is because of a product or equipment blame. Neural systems and routine algorithmic computers are not in rivalry but rather supplement each other. There are assignments are more suited to an algorithmic approach like number juggling operations and errands that are more suited to neural systems. Much more, an expansive number of errands, require
Volume 4, Issue 4, April 2017 pg. 1-12 frameworks that utilization a blend of the two methodologies (ordinarily a routine Computer is utilized to administer the neural system) with a specific end goal to perform at most extreme productivity.
2.1 What is Artificial Neural Network?
An Artificial Neuron Network (ANN) is a computational model based on the structure and functions of biological neural networks. Information that flows through the network affects the structure of the ANN because a neural network changes or learns, in a sense - based on that input and output. Artificial Neural Networks are moderately rough electronic models in light of the neural structure of the brain. The mind essentially gains for a fact. It is characteristic confirmation that a few issues that are past the extent of current computers are without a doubt reasonable by little strength effective bundles. This brain displaying additionally guarantees a less specialized approach to create machine arrangements. This new way to deal with processing additionally gives a more elegant corruption amid framework over-burden than its more conventional partners. These organically roused techniques for processing are thought to be the following significant progression in the registering business.
Indeed, even basic creature brains are fit for capacities that are presently unimaginable for Computers. Computers do repetition things well, such as keeping records or performing complex math. Yet, Computers experience difficulty perceiving even basic examples considerably less summing up those examples of the past into activities without bounds. Presently, progresses in organic research guarantee an underlying comprehension of the regular intuition instrument. This examination demonstrates that brains store data as examples. Some of these examples are exceptionally entangled and permit us the capacity to perceive singular appearances from a wide range of edges. This procedure of putting away data as examples, using those examples, and after those tackling issues envelops another field in processing. This field, as said some time recently, does not use conventional programming but rather includes the production of enormously parallel systems and the preparation of those systems to take care of particular issues. This field additionally uses words altogether different from customary figuring, words like carry on, respond, self-sort out, learn, sum up, and overlook. At whatever point we discuss a neural system, we ought to all the more prevalently say ―Artificial Neural Network (ANN), and ANN are Computers whose engineering is designed according to the brain. They normally comprise of many basic preparing units which are wired together in an unpredictable correspondence arrange. Every unit or hub is astreamlined model of genuine neuron which sends off another flag or flames on the off chance that it gets an adequately solid Input motion from alternate hubs to which it is associated.
Fig 2: Artificial Neuron & Multilayer artificial network
Traditionally neural network was used to refer as network or circuit of biological neurons, but modern usage of the term often refers to ANN. ANN is mathematical model or computational model, an information processing paradigm i.e. inspired by the way biological nervous system, such as brain information system. ANN is made up of interconnecting artificial neurons which are programmed like to mimic the properties of m biological neurons. These neurons working in unison to solve specific problems. ANN is configured for solving artificial intelligence problems without creating a model of real biological system.
ANN is used for speech recognition, image analysis, adaptive control etc. These applications are done through a learning process, like learning in biological system, which involves the adjustment between neurons through synaptic connection. Same happen in the ANN.
Volume 4, Issue 4, April 2017 pg. 1-12 2.2. Why Artificial Neural Network?
The long evolution has given many best and excellent characteristics to brain of human being which are not present in modern computers which are:
Massive Parallelism
Distributed representation and computation
Adaptability
Learning Ability
Generalization Ability
Inherent Contextual Information Processing
Fault Tolerance
Love Energy Consumption
2.3 Differences between Modern Computers and Biological Neural System Modern Computers:-
Contain one or few Processors which are high speed but complex.
Having Localized Memory separate from processor.
Computing is done with stored programs in a sequential and centralized manner.
In terms of reliability it is very vulnerable.
The Operating Environment is well defined and well constrained.
Biological Neural system:-
Contains a large number of processor which have low speed but simple in structure.
Having Distributed Memory but integrated into processor.
Computing is done with self-learning in a parallel and distributed manner.
In terms of reliability it is robust.
The operating environment is poorly defined and unconstrained.
2.4 Characteristics of Artificial Neural Network
Fundamentally Computers are great in counts that essentially takes inputs handle then and after that gives the outcome on the premise of computations which are done at specific Algorithm which are customized in the product's yet ANN enhance their own particular guidelines, the more choices they make, the better choices may get to be. The Characteristics are fundamentally those which ought to be available in smart System like robots and other Artificial Intelligence Based Applications. There are six characteristics of Artificial Neural Network which are basic and important for this technology which are showed with the help of diagram:-
Volume 4, Issue 4, April 2017 pg. 1-12
Fig 5: Characteristics
III. STRUCTURE OF THE ARTIFICIAL NATURAL WORK
The Network Structure of ANN ought to be straightforward and simple. There are essentially two sorts of structures intermittent and non-repetitive structure. The Recurrent Structure is otherwise called Auto acquainted or Feedback Network and the Non Recurrent Structure is otherwise called Associative or sustains forward Network. In Feed forward Network, the flag go in one way just however in Feedback Network, the flag go in both the bearings by presenting circles in the system.The Figures are given below which shows the direction of signals in both the network structures Feed forward and feedback.
Fig 6: Feed Forward Network & Feed Back Network
3.1. Parallel Processing Ability:-
ANN is just acquainting with expand the idea of parallel preparing in the computer field. Parallel Processing is finished by the human body in human neurons are extremely unpredictable yet by applying essential and basic parallel preparing procedures we execute it in ANN like Matrix and some lattice estimations.
3.2 Distributed Memory:-
ANN is very huge system so single place memory or centralized memory cannot fulfill the need of ANN system so in this condition we have to store data in weight lattice which is type of long haul memory since data is put away as examples all through the system structure
3.3 Fault Tolerance Ability:-
ANN is a very complex system so it is necessary that it should be a fault tolerant. Because if any part becomes fail it will not affect the system as much but if the all parts fails at the same time the system will fails completely.
Volume 4, Issue 4, April 2017 pg. 1-12 3.4 Collective Solution:-
ANN is an interconnected system the output of a system is a collective output of various input so the result is summation of all the outputs which comes after processing various inputs. The Partial answer is worthless for any user in the ANN System.
3.5. Learning Ability:-
In ANN most of the learning rules are used to develop models of processes, while adopting the network to the changing environment and discovering useful knowledge. These Learning methods are Supervised, Unsupervised and Reinforcement Learning.
3.5. Activation Function
Activation Functions are basically the transfer function which is output from a artificial neuron and it send signals to the other artificial neuron. There are four forms of Activation Functions Threshold, Piecewise Linear, Sigmoid and Gaussian all are different from each other.
IV. NETWORK ARCHITECTURES
There are further divisions of Feedback and Feed Forward Network architecture which are shown in below Figure
Fig 8: Taxonomy of Network Architecture
4.1Working of Artificial Neural Network:
The other parts of the ―art of using neural networks revolve around the myriad of ways these individual neurons can be clustered together. This clustering occurs in the human mind in such a way that information can be processed in a dynamic, interactive, and self-organizing way. Biologically, neural networks are constructed in a three-dimensional world from microscopic components. These neurons seem capable of nearly unrestricted interconnections. That is not true of any proposed, or existing, man-made network. Integrated circuits, using current technology, are two-dimensional devices with a limited number of layers for interconnection. This physical reality restrains the types, and scope, of artificial neural networks that can be implemented in silicon. Currently, neural networks are the simple clustering of the primitive artificial neurons. This clustering occurs by creating layers which are then connected to one another. How these layers connect is the other part of t he
"art" of engineering networks to resolve real world problems.
Volume 4, Issue 4, April 2017 pg. 1-12
Figure 9:- A Simple Neural Network Diagram &Functions of an Artificial Neuron
Essentially, all simulated neural systems have a comparative structure or topology. In that structure a portion of the neurons interfaces to this present reality to get its sources of info. Different neurons give these present reality the system’s yields. This yield may be the specific character that the system believes that it has checked or the specific picture it supposes is being seen.
All whatever remains of the neurons are escaped see. Be that as it may, a neural system is more than a bundle of neurons. Some early analysts attempted to just associate neurons in an arbitrary way, without much achievement. Presently, it is realized t hat even the brains of snails are organized gadgets. One of the least demanding approaches to outline a structure is to make layers of components. It is the gathering of these neurons into layers, the associations between these layers, and the summation and exchange works that involves a working neural system. The general terms used to portray these attributes are normal to all systems. Despite the fact that there are helpful systems which contain just a single layer, or even one component, most applications require systems that contain in any event the three typical sorts of layers - input, covered up, and yield. The layers of info neurons get the information either from information records or straightforwardly from electronic sensors progressively applications. The yield layer sends data specifically to the outside world, to an auxiliary computer handle, or to different gadgets, for example, a mechanical control framework. Between these two layers can be many concealed layers. These inward layers contain a considerable lot of the neurons in different interconnected structures. The sources of info and yields of each of these shrouded neurons just go to different neurons. In many systems every neuron in a concealed layer gets the signs from the majority of the neurons in a layer above it, normally an information layer. After a neuron plays out its capacity it passes its yield to the greater part of the neurons in the layer underneath it, giving an encourage forward way to the yield. (Note: in segment 5 the drawings are turned around, data sources come into the base and yields turn out the top.)
These lines of correspondence starting with one neuron then onto the next are imperative parts of neural systems. They are the paste to the framework. They are the associations which give a variable quality to information. There are two sorts of these associations. One causes the summing instrument of the following neuron to include while alternate causes it to subtract. In more human terms one energizes while alternate hinders. A few systems need a neuron to hinder alternate neurons in a similar layer. This is called parallel restraint. The most widely recognized utilization of this is in the yield layer. For instance in content acknowledgment if the likelihood of a character being a "P" is .85 and the likelihood of the character being a "F" is .65, the system needs to pick the most noteworthy likelihood and repress all the others. It can do that with parallel hindrance. This idea is likewise called rivalry. Another sort of association is input. This is the place the yield of one layer courses back to a past layer. A case of this is appeared in Figure 10.
Volume 4, Issue 4, April 2017 pg. 1-12
Figure 10:- Simple Network with Feedback and Competition.
The way that the neurons are associated with each different significantly affects the operation of the system. In the bigger, more expert programming improvement bundles the client is permitted to include, erase, and control these associations voluntarily. By "tweaking" parameters these associations can be made to either energize or restrain.
V. ARTIFICIAL NEURAL NETWORKING
Once a system has been organized for a specific application, which system is prepared to be prepared? To begin this procedure the underlying weights are picked haphazardly. At that point, the preparation, or learning, starts. There are two ways to deal with preparing - directed and unsupervised. Administered preparing includes an instrument of giving the system the coveted yield either by physically "evaluating" the system's execution or by furnishing the coveted yields with the sources of info. Unsupervised preparing is the place the system needs to comprehend the contributions without outside offer assistance.
The immeasurable heft of systems use directed preparing. Unsupervised preparing is utilized to play out some underlying portrayal on sources of info. Notwithstanding, in the all-out feeling of being really self-learning, it is still only a sparkling guarantee that is not completely comprehended, does not totally work, and along these lines is consigned to the lab.
5.1 Supervised Training:
In directed preparing, both the sources of info and the yields are given. The system then procedures the sources of info and analyzes its subsequent yields against the coveted yields. Mistakes are then proliferated back through the framework, bringing about the framework to modify the weights which control the system. This procedure happens again and again as the weights are constantly changed. The arrangement of information which empowers the preparation is known as the "preparation set."
During the preparation of a system a similar arrangement of information is handled commonly as the association weights are ever refined. The present business arrange improvement bundles give devices to screen how well a counterfeit neural system is uniting on the capacity to anticipate the correct answer. These apparatuses permit the preparation procedure to continue for a considerable length of time, halting just when the framework achieves some factually wanted point, or exactness. Be that as it may, a few systems never learn. This could be on the grounds that the info information does not contain the particular data from which the coveted yield is determined. Arranges additionally don't meet if there is insufficient information to empower finis h learning. In a perfect world, there ought to be sufficient information so that piece of the information can be kept down as a test.
Many layered systems with numerous hubs are fit for remembering information. To screen the system to figure out whether the framework is basically remembering its information in some non-critical way, directed preparing needs to keep down an arrangement of information to be utilized to test the framework after it has experienced its preparation. On the off chance t hat a system basically can't tackle the issue, the planner then needs to survey the information and yields, the quantity of layers, the quantity of components per layer, the associations between the layers, the summation, exchange, and preparing capacities, and even the underlying weights themselves. Those progressions required to make a fruitful system constitute a procedure wherein
Volume 4, Issue 4, April 2017 pg. 1-12 the "workmanship" of neural systems administration happens. Another piece of the planner's innovativeness oversees the principles of preparing. There are numerous laws (calculations) used to execute the versatile input required to alter the weights amid preparing. The most widely recognized strategy is in reverse mistake proliferation, all the more generally known as back- spread. These different learning strategies are investigated in more prominent profundity later in this report. However, preparing is not only a procedure. It includes a "vibe," and cognizant investigation, to guarantee that the system is not over prepared. At first, a simulated neural system designs itself with the general measurable patterns of the information. Afterward, it keeps on finding out about different parts of the information which might be spurious from a general perspective. At the point when at last the framework has been effectively prepared, and no further learning is required, the weights can, if wanted, be "solidified."
In a few frameworks this concluded system is then transformed into equipment with the goal that it can be quick. Different frameworks don't secure themselves however keep on learning while underway utilize.
5.2 Unsupervised, or Adaptive Training:
The other sort of preparing is called unsupervised preparing. In unsupervised preparing, the system is furnished with data sources yet not with sought yields. The framework itself should then choose what highlights it will use to gather the info information. This is regularly alluded to as self-association or adaption. Right now, unsupervised learning is not surely known.
This adaption to nature is the guarantee which would empower sci-fi sorts of robots to consistently learn all alone as they experience new circumstances and new conditions. Life is loaded with circumstances where correct preparing sets don't exist.
Some of these circumstances include military activity where new battle methods and new weapons may be experienced. As a result of this startling perspective to life and the human yearning to be readied, there keeps on being exploration into, and seek after, this field. However, right now, the immeasurable main part of neural system work is in frameworks with directed learning.
Directed learning is accomplishing comes about.
VI. APPLICATION
The various real time application of Artificial Neural Network is as follows:
Function approximation, or regression analysis, including time series prediction and modeling.
Call control- answer an incoming call (speaker-ON) with a wave of the hand while driving.
Classification, including pattern and sequence recognition, novelty detection and sequential decision making.
Skip tracks or control volume on your media player using simple hand motions- lean back, and with no need to shift to the device- control what you watch/ listen to.
Data processing, including filtering, clustering, blind signal separation and compression.
Scroll Web Pages, or within an eBook with simple left and right hand gestures, this is ideal when touching the device is a barrier such as wet hands are wet, with gloves, dirty etc.
Application areas of ANNs include system identification and control (vehicle control, process control), game-playing and decision making (backgammon, chess, racing), pattern recognition (radar systems, face identification, object recognition, etc.), sequence recognition (gesture, speech, handwritten text recognition), medical diagnosis, financial applications, data mining (or knowledge discovery in databases, "KDD").
Another interesting use case is when using the Smartphone as a media hub, a user can dock the device to the TV and watch content from the device- while controlling the content in a touch-free manner from afar.
If your hands are dirty or a person hates smudges, touch-free controls are a benefit.
Volume 4, Issue 4, April 2017 pg. 1-12 VII. ADVANTAGES OF ARTIFICIAL NEURAL NETWORKING
Adaptive learning: An ability to learn how to do tasks based on the data given for training or initial experience.
Self-Organization: An ANN can create its own organization or representation of the information it receives during learning time.
Real Time Operation: ANN computations may be carried out in parallel, and special hardware devices are being designed and manufactured which take advantage of this capability.
Pattern recognition: Pattern recognition is a powerful technique for harnessing the information in the data and generalizing about it. Neural nets learn to recognize the patterns which exist in the data set.
The system is developed through learning rather than programming. Neural nets teach themselves the patterns in the data freeing the analyst for more interesting work.
Neural networks are flexible in a changing environment. Although neural networks may take some time to learn a sudden drastic change they are excellent at adapting to constantly changing information.
Neural networks can build informative models whenever conventional approaches fail. Because neural networks can handle very complex interactions they can easily model data which is too difficult to model with traditional approaches such as inferential statistics or programming logic.
Performance of neural networks is at least as good as classical statistical modeling, and better on most problems. The neural networks build models that are more reflective of the structure of the data in significantly less time.
VIII. LIMITATION OF ARTIFICIAL NEURAL NETWORKING
In this technological era every have Merits and some Demerits in others words there is a Limitation with every system which makes this ANN technology weak in some points. The various Limitations of ANN are:
ANN is not a daily life general purpose problem solver.
There is no structured methodology available in ANN.
There is no single standardized paradigm for ANN development.
The Output Quality of an ANN may be unpredictable.
Many ANN Systems does not describe how they solve problems.
Black box Nature.
Greater computational burden.
Proneness to over fitting.
Empirical nature of model development.
IX. FUTURE WORKS
We talk about the Future work we can only say that we have to develop much more algorithms and other problem solving techniques so that we can remove the limitations of the Artificial Neural Network. And if the Artificial Neural Network concepts combined with the Computational Automata and Fuzzy Logic we will definitely solve some limitations of this excellent technology.
Volume 4, Issue 4, April 2017 pg. 1-12 X. CONCLUSION
The other kind of preparing is called unsupervised preparing. By concentrate fake Neural Network we had presumed that according to as innovation is creating step by step the need of Artificial Intelligence is expanding a direct result of just parallel handling. Parallel Processing is more required in this present time in light of the fact that with the assistance of parallel handling no one but we can spare increasingly time and cash in any business related to PCs and robots. In this paper we examined about the artificial neural system, working of ANN. Additionally preparing periods of an ANN. There are different points of interest of ANN over routine methodologies. Contingent upon the way of the application and the quality of the interior information designs you can by and large anticipate that a system will prepare great. This applies to issues where the connections might be very dynamic or non-direct. ANNs give a scientific other option to customary procedures which are regularly constrained by strict suspicions of typicality, linearity, variable freedom and so on. Since an ANN can catch numerous sorts of connections it permits the client to rapidly and moderately effectively demonstrate marvels which generally may have been extremely troublesome or impossible to clarify something else. Today, neural systems discourses are happening all over. Their guarantee appears to be splendid as nature itself is the verification that this sort of thing works. However, its future, surely the exceptionally key to the entire innovation, lies in equipment improvement. Presently most neural system advancement is basically demonstrating that the vital works.
ACKNOWLEDGEMENT
Author is very thankful to Allah for encouraging the authors to complete the paper successfully. Here, Authors wish to express his sincere admiration to his supervisor for inspiring him to implement this paper perfectly. The authors also would like to acknowledge to contribution for made by the Jagannath University (JnU) authority & who give us an environment where we work more time. In particular, we would like to be grateful all staffs and Liberian, for their collaboration, subsidiary or straight support to complete the entire studies.
References
1. Herve Debar, Monique Becker and Didier Siboni“ A Neural Network Component for an Intrusion Detection System”, Les UlisCedex France, 1992.
2. Ajith Abraham, “Artificial Neural Networks”, Stillwater,OK, USA, 2005.
3. Carlos Gershenson, “Artificial Neural Networks for Beginners”, United kingdom.
4. Anil K Jain, Jianchang Mao and K.M Mohiuddin, “Artificial Neural Networks: A Tutorial”, Michigan State university, 1996.
5. Ugur HALICI, “ Artificial Neural Networks”, Chapter 1, ANKARA.
6. Eldon Y. Li, “ Artificial Neural Networks and their Business Applications”, Taiwan, 1994.
7. Christos Stergiou and DimitriosSiganos, “Neural Networks”.
8. Limitations and Disadvantages of Artificial Neural Network from website http://www.ncbi.nlm.nih.gov/pubmed/8892489.
9. Image of a Neuron form website http://transductions.net/ 2010/02/04/313/neurons/.
10. About Artificial Neural Network from website. http://en.wikipedia.org/wiki/Artificial_neural_network.
11. RC Chakraborty, “Fundamentals of Neural Networks”, myreaders.info/html/artificial_intelligence.html, june 01, 2010.
12. Prof. Leslie Smith, “An Introduction to Neural Networks”, University of Stirling, 1996,98,2001,2003.
13. Prof. Dr. Eduardo Gasca A., “Artificial Neural Networks”, Toluca.
14. KishanMehrotra, Chilukuri K Mohan and Sanjay Ranka “Elements of artificial neural network”, 1996.
15. Weyiu Yi 339229, “Artificial Neural Networks”, 2005.
16. Vincent Cheung and Kevin Cannons, “An Introduction of Neural Networks”, Manitoba, Canada, May 27, 2002.
17. Howard Demuth and Mark Beale, “Neural Network Toolbox”, With the help of metlab, user guide version 4.
18. Girish Kumar Jha, “Artificial Neural Network and its Applications”, IARI New delhi.
19. About Neural Network from website http://en.wikipedia.org / wiki/Neural_network.
20. About Feed Back Network from website http://www.idsia.ch/ ~juergen/rnn.html . 21. Sucharita Gopal, “Artificial Neural Networks for Spatial Data Analysis”, Boston, 1988.
22. Bradshaw, J.A., Carden, K.J., Riordan, D., 1991. Ecological ―Applications Using a Novel Expert System Shell‖. Comp. Appl. Biosci. 7, 79–83.
23. Lippmann, R.P., 1987. An introduction to computing with neural nets. IEEE Accost. Speech Signal Process. Mag., April: 4-22.
24. N. Murata, S. Yoshizawa, and S. Amari, ―Learning curves, model selection and complexity of neural networks,‖ in Advances in Neural Information Processing Systems 5, S. Jose Hanson, J. D. Cowan, and C. Lee Giles, ed. San Mateo, CA: Morgan Kaufmann, 1993, pp. 607-614
Volume 4, Issue 4, April 2017 pg. 1-12
25. Ms. Sonali. B. Maind, Ms. PriyankaWankar ,Research Paper on Basic of Artificial Neural Network, International Journal on Recent and Innovation Trends in Computing and Communication ISSN: 2321-8169 Volume: 2 Issue: 1 96 – 100.
AUTHOR(S)PROFILE
Md Tanjil Sarker, is the corresponding author of this paper. He was born 1991 in Rangpur District in Bangladesh. He successfully completed Bachelor Degree from Bangladesh University in the department of EEE, and He had also completed MBA degree from Business Faculty of Bangladesh University. Now he is studying post graduate degree in the Department of CSE, Jagannath University Dhaka Bangladesh.
He conducted many research works in the relevant field such as Design, Inspection and Implementation of Solar PV Driven Smart& Automated Irrigation Systems, Electricity Load Calculative Method of an Inaccessible area of Bangladesh. Now he is working as an Engineer in the organization of Bangladesh Research and education network (BdREN).
Sanjida Noor,born in Rajshahi in Bangladesh. She accomplished her Bachelor degree in the area of Computer Science & Engineering from University Information Technology & Sciences (UITS), Bangladesh. She conducted many research works in the relevant field. Now she is working as an Engineer in a renowned group of Company in Bangladesh.
Dr. Uzzal Kumar Acharjee, Chairman & Associate Professor, Dept. of Computer Science &
Engineering, Jagannath University, Dhaka 1100, Bangladesh. For More Details please visit:
http://jnu.ac.bd/profile/portal/web/1.html
