2026-05-12T23:37:46.984Z http://ezaposleni.singidunum.ac.rs/rest/sciNaucniRezultati/oai

ezaposleni.singidunum.ac.rs/rest/sciNaucniRezultati/oai:3:11368 2025-04-08T00:11:28Z 3

Waste classification using convolutional neural networks tuned by modified metaheuristics algorithm 2025 http://ezaposleni.singidunum.ac.rs/rest/sciNaucniRezultati/oai/record/3/11368 https://www.sciencedirect.com/science/article/abs/pii/B9780443273742000091 A. Tasic L. Jovanovic M. Popovic N. Bacanin M. Zivkovic M. Dobrojevic Waste classification is a crucial process that aids in managing and mitigating the environmental and health impacts related to waste. It helps protect the environment, as toxic waste may contaminate soil, water, and air, which in turn can lead to serious damage to the ecosystems and human society. Implementing proper classification techniques can aid in the development of appropriate disposal methods that can prevent environmental deterioration. It also helps in waste management efficiency, as waste can be categorized into separated groups with respect to the characteristics, including toxicity, flammability, and biodegradability, allowing selection of the most appropriate handling and disposal options. The approach presented in this paper relies on the application of convolutional neural networks to perform the waste classification task. Every deep learning method must be optimized for the particular problem. Emphasizing the significance of proper hyperparameter selection, the research evaluates various contemporary optimizers through an iterative selection process. Additionally, a modified version of the crayfish optimization algorithm (COA) is introduced, tailored specifically for the study’s requirements. The evaluation of the proposed approach on real-world data consistently reveals that models configured by the introduced classifier surpass an accuracy of 87.98%. This underscores the significant potential of the approach in addressing urgent waste management challenges in real-world scenarios. bookPart Elsevier 237 261 10.1016/B978-0-443-27374-2.00009-1 Chapter in Harnessing Automation and Machine Learning for Resource Recovery and Value Creation