CS 631- Research Trends In AI
Report 2
Submitted by: Group 2
Applications of Artificial Neural Networks
Introduction
Artificial Neural Networks (ANN) attempt to simulate the functioning of
biological neural systems on a simplified level. Nodes in the network represent
neurons in a biological system. Each node has some inputs and output
connection lines and can be in one of the two possible states, active or inactive.
The output lines of a node are connected to the input lines of other nodes. Each
node in a layer is linked by a weighted connection to every node in the layer that
precedes it. The weight is assigned to each connection to represent the strength
of that connection. A node computes its state as a function of the weights of the
input connections to adjacent nodes and the states of the adjacent nodes. The
ANN may be organized into an input layer, an output layer and a number of
intermediate layers. An input pattern is applied to the ANN by activating certain
nodes, and an output pattern is produced as a function of the weights of the
connections. The ANN learns by comparing the resulting output pattern to the
correct output and adjusting the weights of the connections accordingly. The
hidden and output layers’ nodes hold additional information on the connections,
or weights, between nodes and calculate their states from this information
during their calculation iterations. If the node’s error is positive this weight is
then redistributed to the active input lines, thus strengthening the connections to
the active nodes below. If the node’s error is negative, the weight is redistributed
to the inactive input lines, reducing the strength of the connections to the active
nodes below. Hence, the weights of the output and hidden layers are adjusted to
compensate for error.
Applications
Neural networks have been successfully applied to a broad spectrum of dataintensive applications.
Stock Market Prediction – Helps predict the future movement of security using
historical data of that security.
Data Mining – Uses some variables or fields in a database to predict unknown or
future values of other variables of interest.
Medical Diagnosis – Assists doctors with their diagnosis by analyzing reported
symptoms.
CS 631- Research Trends In AI
Report 2
Submitted by: Group 2
Detection and Evaluation of Medical Phenomena – Detects epileptic attacks,
estimates prostate tumor size and detects patient breathing abnormalities when a
patient is under anesthesia, etc.
Patient's Length of Stay Forecasts – Forecasts which patients remain for a
specified number of days.
Sales Forecasting – Predicts future sales based on historical information about
previous
marketing
and
sales
activities.
Targeted Marketing – Reduces costs by targeting a particular marketing
campaign to the group of people which have the highest response rate. Avoid
wasting
money
on
unlikely
targets.
Service Usage Forecasting – Forecasts the number of service calls, customer
transactions, customer arrivals, reservations or restaurant covers (patrons) in
order to effectively schedule enough staff to handle the workload.
Retail Margins Forecasting – Forecast the behavior of margins in the future to
determine the effects of price changes at one level on returns at the other.
Process Control – Determine the best control settings for a plant. Complex
physical and chemical processes that may involve interaction of numerous
(possibly unknown) mathematical formulas can be modeled heuristically using a
neural network.
Quality Control – Predicts the quality of plastics, paper, and other raw materials;
machinery defect diagnosis; diesel knock testing, tire testing, beer testing.
Retail Inventories Optimization – Forecasts optimal stock level that can meet
customer needs, reduce waste and lessen storage; predict the demand based on
previous buyers' activity.
Scheduling Optimization – Predicts demand to schedule buses, airplanes, and
elevators.
Managerial Decision Making – Selects the best decision option using the
classification capabilities of neural network.
CS 631- Research Trends In AI
Report 2
Submitted by: Group 2
Cash Flow Forecasting – Maximizes the use of resources with more accurate
cash flow forecasts.
Employee Selection and Hiring – Predicts on which job an applicant will
achieve the best job performance.
Employee Retention – Identifies potential employees who are likely to stay with
the organization for a significant amount of time based on data about an
applicant.
Staff Scheduling – Predicts staff requirements for restaurants, retail stores,
police stations, banks, etc.
OUTPUT
HIDDEN
INPUT LAYER
ACTIVE
INACTIVE
Fig. 1. An Artificial Neural Network
CS 631- Research Trends In AI
Report 2
Submitted by: Group 2
References


http://www.alyuda.com/products/forecaster/neural-network-applications.htm
http://web.syr.edu/~gscott/research/neuralnet.html