RT Conference Proceedings
T1 End-to-end Incremental Learning
A1 Castro Payán, Francisco Manuel
A1 Marín-Jiménez, Manuel J.
A1 Guil-Mata, Nicolás
A1 Schmid, Cordelia
A1 Alahari, Karteek
K1 Redes neuronales (Informática)
AB Although deep learning approaches have stood out in recent years due to their state-of-the-art results, they continue to suffer from  (catastrophic forgetting), a dramatic decrease in overall performance when training with new classes added incrementally. This is due to current neural network architectures requiring the entire dataset, consisting of all the samples from the old as well as the new classes, to update the model---a requirement that becomes easily unsustainable as the number of classes grows. We address this issue with our approach to learn deep neural networks incrementally, using new data and only a small exemplar set corresponding to samples from the old classes. This is based on a loss composed of a distillation measure to retain the knowledge acquired from the old classes, and a cross-entropy loss to learn the new classes. Our incremental training is achieved while keeping the entire framework end-to-end, i.e., learning the data representation and the classifier jointly, unlike recent methods with no such guarantees.
YR 2018
FD 2018-07-06
LK https://hdl.handle.net/10630/16158
UL https://hdl.handle.net/10630/16158
LA eng
NO This work has been funded by project TIC-1692 (Junta de Andalucía), TIN2016-80920R (Spanish Ministry of Science and Technology) and  Universidad de Málaga. Campus de Excelencia Internacional Andalucía Tech.
DS RIUMA. Repositorio Institucional de la Universidad de Málaga
RD 21 ene 2026