ETD

• automation
• continual
• industry
• learning
• sustainable

The objective of this thesis is to evaluate the effectiveness of Continual Learning, in conjunction with Knowledge Distillation, in industrial-like environments. Given that these are key techniques for an effective Industry 4.0 transition, we propose a blueprint for integrating these sustainable Deep Learning approaches within manufacturing processes. Finally, we assess the performance of some models trained according to this new protocol, on data streams that are qualitatively consistent with those obtainable from a real production line.