Browse Prior Art Database

Improved Multimedia Response using Neural Net Prediction

IP.com Disclosure Number: IPCOM000113834D
Original Publication Date: 1994-Oct-01
Included in the Prior Art Database: 2005-Mar-27
Document File: 2 page(s) / 54K

Publishing Venue

IBM

Related People

Cragun, BJ: AUTHOR [+3]

Abstract

A method for improving response times of interactive multimedia applications is disclosed. A neural network is trained to predict user selection based on previous user selection and time-based information. Multimedia is then prepared based on prediction.

This text was extracted from an ASCII text file.
This is the abbreviated version, containing approximately 62% of the total text.

Improved Multimedia Response using Neural Net Prediction

      A method for improving response times of interactive multimedia
applications is disclosed.  A neural network is trained to predict
user selection based on previous user selection and time-based
information.  Multimedia is then prepared based on prediction.

      An interactive multimedia application is given access to a
neural network.  As the application comes to a logic branch dependant
on user input, it provides the neural network with status
information.  The neural network returns a prediction of the next
user selection.  When the maximum network output value is above a
specified threshold value, the response or branch represented by the
output is used as the prediction.  Based on this prediction, the
application prepares the needed multimedia objects.  For example,
digital objects are pre-loaded in memory and videodisc objects are
pre-searched.

      When the neural network correctly predicts the next user
selection, the user experiences improved performance.  If the
prediction is incorrect, the performance corresponds to the actual
time of loading the users actual response, which is not as good, and
considered an average response time.

      Inputs to the neural network are:  previous 10 responses, time
of day, day of week, holiday designation, time taken for last
response, and elapsed time from first response.  The outputs of the
neural network correspond to possible branches...