Hybrid deep learning model for accurate classification of solid waste in the society

Due to the increasing initiatives for urbanization and the development of smart cities, waste generation, segregation, and its management have become fundamental tasks. To provide efficient planning for waste management and its processes, such as collection, sorting, recycling, and disposal, recently, machine learning (ML) approaches have been utilized to assist the authorities. However, the identification of the best ML approach for the prediction of waste is a challenging effort. Finding adequate waste litter measurement is necessary for the ecological characteristics to improve over time. The waste from the trash may divide into organic and recycling types. In this paper, the optimized hybrid deep learning model has been developed for waste classification. This proposed work takes advantage of (i) data collection and preprocessing (ii) feature extraction using CNN (AlexNet) (iii) waste prediction from the urban cities' wastes using DBN, and (iv) hyperparameter optimization using Optuna. This model obtained an R2 score of 0.94, MPE 0.02 than other state-of-the-art approaches. Compared to the individual learners model, this proposed optimized hybrid deep learning model boosts the performance to predict waste generation and classify it with increased accuracy.