Preliminary communication
Online service for accessible machine learning of prediction models
Ivan Zatezalo
orcid.org/0000-0001-9245-2255
; Filozofski fakultet Sveučilišta u Zagrebu, Zagreb, Hrvatska
Ivan Dunđer
; Filozofski fakultet Sveučilišta u Zagrebu, Zagreb, Hrvatska
Abstract
The field of machine learning is getting more and more advanced every day.
Many use this technology in order to start their projects or better their current applications.
Due to such a great interest, the use of mathematical algorithms required to build artificial
intelligence has been simplified through various frameworks that allow seamless
implementation. One framework by the name Teachable Machine is available as an online
service. Its ease of use and simplicity allows anyone to develop machine learning models that
can then be applied for personal use or projects. Despite this, such a framework still requires
specialized knowledge in order to make an effective and accurate prediction model. This
paper aims to test the limits of this service by developing models of different complexity. One
model will be tested using the upload feature present on the service, whereas the other one
will be implemented within a mobile application and tested on real-world data. The results
gathered from this testing will then give an insight into the performance and quality of the
aforementioned online service.
Keywords
machine learning; image classification; Teachable Machine; model accuracy; TensorFlow; information and communication sciences
Hrčak ID:
267302
URI
Publication date:
15.12.2021.
Visits: 815 *