| Abstract |
Deep clustering is a deep learning model, practically a neural network that incorporates clustering, used to handle high-dimensional and complex data. It has shown a good performance, as it has ability to Manage Complex and high-dimensional data, Automatic Feature Extraction. It is applied in many fields as Image Segmentation and Image Clustering, NLP, Bioinformatics and Healthcare, Urban Planning, Smart Cities and many other areas . In our approach we aim to use deep clustering concept to determine the optimal solutions of a multi objective optimization problem by studying and assessing dominance between its feasible solutions. It is a deep clustering-based method named Dominance Assessment Deep Clustering Based Approach (DeepDOM) composed of three layers, namely, Input layer is dedicated to select the Input data, which is selected by using SVM. The hidden layers are dedicated to, first, clustering assignment layer, which is done by using a binary order relation with clusters number not already known, but counted during the learning. Second, dominance assessment layer which is dedicated to determine the minimum element within each cluster. The final /output layer is dedicated to extract the optimal solutions. © The Author(s), under exclusive license to Springer Nature Switzerland AG 2025. |
