RESUMEN
The staple of human intelligence is the capability of acquiring knowledge in a continuous fashion. In stark contrast, Deep Networks forget catastrophically and, for this reason, the sub-field of Class-Incremental Continual Learning fosters methods that learn a sequence of tasks incrementally, blending sequentially-gained knowledge into a comprehensive prediction. This work aims at assessing and overcoming the pitfalls of our previous proposal Dark Experience Replay (DER), a simple and effective approach that combines rehearsal and Knowledge Distillation. Inspired by the way our minds constantly rewrite past recollections and set expectations for the future, we endow our model with the abilities to i) revise its replay memory to welcome novel information regarding past data ii) pave the way for learning yet unseen classes. We show that the application of these strategies leads to remarkable improvements; indeed, the resulting method - termed eXtended-DER (X-DER) - outperforms the state of the art on both standard benchmarks (such as CIFAR-100 and miniImageNet) and a novel one here introduced. To gain a better understanding, we further provide extensive ablation studies that corroborate and extend the findings of our previous research (e.g., the value of Knowledge Distillation and flatter minima in continual learning setups). We make our results fully reproducible; the codebase is available at https://github.com/aimagelab/mammoth.
RESUMEN
Foodservice is estimated to produce 12% of the total food waste in Europe, and it is a major target for policies against food waste. The amount of food lost during foodservice operations has been assessed in the literature as a figure ranging from 10% to 41% of the quantity prepared in the kitchen, either as non-served food or as plate waste. In this paper, a systematization of the current initiatives against food waste in EU school foodservice is provided. This background is used as a base to discuss the results of a direct assessment of food waste conducted in 78 primary schools in Italy, where 28.6% of the food prepared was not consumed by the diners. Part of it was saved for reuse, while the rest was disposed and treated as organic waste or, to a lesser extent, as unsorted waste. The flows of food waste, represented by a Sankey diagram, show that some actions may be implemented in order to save more food from disposal; for example, implementing donation programmes for non-served food or using doggy bags to avoid the disposal of plate waste. A greater effort shall be put on preventive actions, aimed at avoiding the generation of food waste; in this sense regular monitoring at schools may act as a first preventive measure as it can increase the awareness of students, teachers and foodservice staff over the issue of food waste.
