<![CDATA[Waste is a serious issue facing the world today, with increasing human activity and global economic growth. One important step in waste management is the classification process, which aims to separate types of waste based on their characteristics so they can be recycled, processed, or disposed of properly. Previous research has shown that Convolutional Neural Networks (CNN) are effective algorithms for multi-class classification. Therefore, this study develops an optimized CNN model for automatic waste classification, with a primary focus on improving accuracy through modifications to the CNN architecture. The dataset used consists of 17,366 waste images from various sources, which are then divided into training and testing data after undergoing preprocessing to ensure good data quality before training the model. However, one of the main challenges in developing a CNN model for multi-class classification is the risk of difficulty in learning class features, especially when the model is faced with too many classes. To address this issue, this study implements a strategy by adding convolutional layers to the CNN architecture. This method aims to deepen the network to capture more complex features from the given data, which in turn can improve the model's generalization to new data. Evaluation results show that the modified CNN model achieved a training accuracy of 88% after 40 epochs, with a testing accuracy of around 83%. This research not only contributes to the development of more advanced automatic waste classification technology but also provides a strong foundation for further research in this field. With increased waste management effectiveness, it is hoped to have a positive impact on the environment and public health as a whole..]]>
