Incremental zero-shot learning based on attributes for image classification

Xue, Nan; Wang, Yi; Fan, Xin; Min, Maomao

doi:10.1109/icip.2017.8296401

Cited by 5 publications

(1 citation statement)

References 20 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Still, the proposed method improves both base methods in both datasets significantly. The setting proposed is different than the -markedly less challenging -setting appearing in Kankuekul et al [19] and Xue et al [40], where the set of unseen classes is fixed, and only more seen classes are added incrementally, without any sample limitations given by rehearsal memory.…”

Section: Main Experimentsmentioning

confidence: 99%

Insights from the Future for Continual Learning

Douillard¹,

Valle²,

Ollion³

et al. 2020

Preprint

View full text Add to dashboard Cite

Continual learning aims to learn tasks sequentially, with (often severe) constraints on the storage of old learning samples, without suffering from catastrophic forgetting. In this work, we propose prescient continual learning, a novel experimental setting, to incorporate existing information about the classes, prior to any training data. Usually, each task in a traditional continual learning setting evaluates the model on present and past classes, the latter with a limited number of training samples. Our setting adds future classes, with no training samples at all. We introduce Ghost Model, a representation-learning-based model for continual learning using ideas from zero-shot learning. A generative model of the representation space in concert with a careful adjustment of the losses allows us to exploit insights from future classes to constraint the spatial arrangement of the past and current classes. Quantitative results on the AwA2 and aP&Y datasets and detailed visualizations showcase the interest of this new setting and the method we propose to address it.

show abstract

Section: Main Experimentsmentioning

confidence: 99%