Abstract:
In our time, the amount of information and tweets are increasing on Twitter. Unfortunately, we
found that Twitter is a popular place for spammers, which share unwanted messages that may contain malicious software, advertisements, or links that contain malicious sites. As a means of avoiding
text-based filters, spammers inject spam text onto images, a process known as image spam. so. How
can we detect these images and know the unwanted messages from it? What are the possible algorithms to detect it ? This is what we will address in this research. In our thesis, we introduce Some
Learning techniques used to classify images as spam or ham and bio-inspired algorithm which used to
optimize the problem, at the experimental level we design convolutional neural network architectures
using the particle swarm optimization algorithm in order to find the optimal network architecture of
convolutional neural networks
