Abstract
Searching for research titles and abstracts is made easy with these
keywords. Artificial Intelligence (AI) technology is currently
developing very rapidly, there are various applications of AI that we
can find in everyday life around us without realizing it. AI technology
now allows us to work with computers more easily, just as we can know
the type of cat breed and other information. There is deep learning that
works by imitating the human brain or artificial neural networks to
enhance current machine-learning capabilities. Deep learning can
recognize and classify image categories. This study aims to determine
the optimal optimizer in the classification of cat breeds. With the
classification of cat breeds, cat keepers can find out the type of cat
breed so they can find out how to care for it, the activities, and the
personality possessed by the cat. The use of algorithm method used in
this study uses the CNN algorithm with the NASNetMobile architecture.
The dataset contains 840 images which are divided into 4 classes and
divided into 588 training data, 168 testing data, and 84 validation
data. for the RMSprop optimizer with a learning rate of 0.0001 to get an
accuracy of 89.88%, this result is the highest among the others.
Meanwhile, the SGD optimizer gets an accuracy of 78.57 & this result
is the lowest. So it can be concluded that the architecture and
optimizer are very important and influential in improving the
performance of the model.