A binary particle swarm optimization-based pruning approach for environmentally sustainable and robust CNNs

Jihene Tmamna, Rahma Fourati, Emna Ben Ayed, Leandro A. Passos, João P. Papa, Mounir Ben Ayed, Amir Hussain

Abstract: Highlights•Layer pruning is regarded as a constrained single-objective optimization problem.•A masked-bit strategy is proposed for reducing the search space.•A modified BPSO initialization method based on the L1 norm is proposed.•Carbon footprint evaluation is handled on varied pre-trained models.•Robustness of the pruned model is evaluated using adversarial attacks.