Deck 6: Artificial Neural Networks for Data Mining

Question 1

Supervised learning uses a set of inputs for which the desired outputs are known.For example,a dataset of loan applications with the success or failure of borrowers to repay their loans has a set of input parameters and known outputs.

Answer 1

True

Question 2

Weights are crucial in network information processing because they store learned patterns of information; and it is through repeated adjustments of these weights that a network learns.

Answer 2

True

Question 3

Neural network models are designed as exact replicas of how the human brain actually functions.

Answer 3

False

Question 4

Implementation of an ANN often requires interfaces with other computer-based information systems and user training.

Question 5

Parallel processing resembles the way the brain and conventional computing works.

Question 6

Typically,the input-output transformation process at the individual neuron level is performed in a linear fashion.

Question 7

Several types of data,such as text,pictures,and voice,can be used as inputs in network information processing.

Question 8

In neural network,larger data sets increase processing time during training but improve the accuracy of the training and often lead to faster convergence to a good set of weights.

Question 9

Training of artificial neural networks is an iterative process and the iteration continues until the error sum is converged to below a preset accep level.

Question 10

In network information processing,each input corresponds to one or two attributes.

Question 11

In general,ANN are sui for problems whose inputs are both categorical and numeric,and where the relationships between inputs and outputs are linear or the input data are normally distributed.

Question 12

Minotaur was implemented to prevent fraud.In the first 3 months following installation of Minotaur,the average fraud loss per case was reduced by 40 percent.

Question 13

A general Hopfield network is a single large layer of neurons with total interconnectivity.

Question 14

The output of a network contains the solution to a problem.

Question 15

Each ANN is composed of a collection of neurons that are grouped into three types of layers: input,intermediate (or hidden),and output.

Question 16

A threshold value is a hurdle value for the output of a neuron to trigger the next level of neurons.If an output value is larger than the threshold value,it will not be passed to the next level of neurons.

Question 17

Neural networks have been used in finance,marketing,manufacturing,operations,and information systems and in many business applications for pattern recognition,forecasting,prediction,and classification.

Question 18

The human brain is composed of special cells called neurons.Neural networks are composed of interconnected processing elements called artificial neurons.

Question 19

The processing elements (PE)of an ANN are artificial neurons,which receive inputs,process them,and deliver outputs.

Question 20

The information processing in neural networks makes it attractive for solving complex problems.

Question 21

Sigmoid function is an S-shaped transfer function in the range of 0 to 1 and is also a useful ________ transfer function.

A) integer
B) binary
C) linear
D) nonlinear

Question 22

The way that information is processed by the neural network is a function of its ________.

A) composition
B) formation
C) structure
D) makeup

Question 23

________ is the most commonly used network paradigm.

A) Parallel processing
B) Processing element
C) Minotaur
D) Backpropagation

Question 24

Pioneers McCulloch and Pitts built their neural network model using a large number of interconnected ________ artificial neurons.

A) dual
B) binary
C) singular
D) serial

Question 25

An artificial neural network is composed of many interconnected ________.

A) artificial units
B) artificial cells
C) artificial neurons
D) artificial atoms

Question 26

There are about 50 to 150 billion neurons in the human brain and these neurons are partitioned into groups called ________.

A) teams
B) sects
C) groups
D) networks

Question 27

The ways neurons are organized are referred to as ________.

A) topologies
B) contour
C) formation
D) configuration

Question 28

Which of the following procedure is used to break datasets into different pairs of training and testing sets?

A) resampling
B) sampling
C) trial and error
D) random

Question 29

The summation function computes the ________ sums of all the input elements entering each processing element.

A) weighted
B) averaged
C) total
D) aggregated

Question 30

The backpropagation learning algorithm is an iterative ________ technique designed to minimize an error function between the actual output of the network and its desired output,as specified in the training set of data.

A) differential equation
B) binary search
C) gradient-descent
D) exact search

Question 31

In single,hidden-layer structured neural network,this hidden layer converts inputs into a ________ combination.

A) continuous
B) linear
C) nonlinear
D) nonstop

Question 32

Most implementations of the learning process in neural network include a counterbalancing parameter called ________ to provide a balance to the learning rate.

A) power
B) energy
C) force
D) momentum

Question 33

Which of the following is not a consideration in selecting a neural network structure?

A) Selection of a topology
B) Determination of input nodes
C) Determination of output nodes
D) Determination of weighting functions

Question 34

The connection weights express the ________ of the input data.

A) significance
B) value
C) mathematical value
D) worth

Question 35

Which of the following is the reason why neural networks have been applied in business classification problems?

A) Able to learn the data
B) Able to learn the models' nonparametric nature
C) Its ability to generalize
D) All of the above

Question 36

Learning algorithms specify the ________ by which a neural network learns the underlying relationship between input and outputs.

A) process
B) method
C) route
D) direction

Question 37

The output of neurons can be the final result or it can be ________ to other neurons.

A) sources
B) contributions
C) keys
D) inputs

Question 38

Which of the following is a trait of an artificial neural network?

A) Fault tolerance
B) Duplicate cell
C) Self-repaired
D) Memory less

Question 39

ANN can also be used as simple biological models to test ________ about biological neuronal information processing.

A) hypotheses
B) assumptions
C) theory
D) proposition

Question 40

Because of their ability to capture and represent highly complex relationships,a new and prosperous area of application for neural networks is in the field of ________.

A) health care and medicine
B) transportation and distribution
C) security
D) financial planning

Question 41

________ refers to a pattern recognition methodology for machine learning.

Question 42

In ________,the network is self-organizing; that is,it organizes itself internally so that each hidden processing element responds strategically to a different set of input stimuli.

Question 43

________ is the central processing portion of the biological neuron.

Question 44

A neural network is composed of processing elements organized in different ways to form the network's ________.

Question 45

A ________ is able to increase or decrease the strength of the connection from neuron to neuron.

Question 46

A(n)________ network represents a brain metaphor of information processing.

Question 47

________ value is a hurdle value for the output of a neuron to trigger the next level of neurons.

Question 48

In neural network,it is through repeated adjustments of ________ that the network learns.

Question 49

One popular approach,known as the feedforward-backpropagation paradigm,in organizing neurons,does not allow any ________ linkage.

Question 50

Each biological neuron possesses axons and ________,finger-like projections that enable the neuron to communicate with its neighboring neurons by transmitting and receiving electrical and chemical signals.

Question 51

There are several neural network paradigms,and one of the easiest ways to differentiate between the various models is on the basis of how these models structurally ________ the human brain.

Question 52

________ testing is used to comparing test results to historical results.

Question 53

A ________ is a layer of neurons that takes input from the previous layer and converts those inputs into outputs for further processing.

Question 54

The early neural network model is called ________.

Question 55

In a neural network,the knowledge is stored in the ________ associated with each connection between two neurons.

Question 56

When information is processed,many of the processing elements of neural network perform their computations at the same time,which is called ________ processing.

Question 57

The purpose of the neural network is to compute the ________ of the output.

Question 58

The ________ is the training procedure used by an artificial neural network.

Question 59

ANN stands for ________.

Question 60

The artificial neurons receive the information from external input stimuli,perform a ________ on the inputs,and then pass on external outputs.

Question 61

List the procedures of the learning algorithm.

Question 62

List the relationships between biological and artificial networks.

Question 63

Explain supervised and unsupervised learning modes of neural networks.

Question 64

List the usual process of general learning in neural network.

Question 65

Explain threshold value and its role in the network.

Question 66

Describe the five major concepts / components of neural network information processing.

Question 67

Briefly describe backpropagation.