LAMP

Category: Publications

  • Disentanglement of Color and Shape Representations for Continual Learning

    David Berga, Marc Masana, Joost Van de Weijer Read Full Paper → We hypothesize that disentangled feature representations suffer less from catastrophic forgetting. As a case study we perform explicit disentanglement of color and shape, by adjusting the network architecture. We tested classification accuracy and forgetting in a task-incremental setting with Oxford-102 Flowers dataset. We combine our […]

  • On Class Orderings for Incremental Learning

    Marc Masana, Bartłomiej Twardowski, Joost van de Weijer Read Full Paper → The influence of class orderings in the evaluation of incremental learning has received very little attention. In this paper, we investigate the impact of class orderings for incrementally learned classifiers. We propose a method to compute various orderings for a dataset. The orderings are derived […]

  • RATT: Recurrent Attention to Transient Tasks for Continual Image Captioning

    Riccardo Del Chiaro, Bartłomiej Twardowski, Andrew D. Bagdanov, Joost van de Weijer Read Full Paper → Research on continual learning has led to a variety of approaches to mitigating catastrophic forgetting in feed-forward classification networks. Until now surprisingly little attention has been focused on continual learning of recurrent models applied to problems like image captioning. In this paper […]

  • DeepI2I: Enabling Deep Hierarchical Image-to-Image Translation by Transferring from GANs

    Yaxing Wang, Lu Yu, Joost van de Weijer Read Full Paper → Image-to-image translation has recently achieved remarkable results. But despite current success, it suffers from inferior performance when translations between classes require large shape changes. We attribute this to the high-resolution bottlenecks which are used by current state-of-the-art image-to-image methods. Therefore, in this work, we propose […]

  • Mix and match networks: cross-modal alignment for zero-pair image-to-image translation

    Yaxing Wang, Luis Herranz, Joost van de Weijer Read Full Paper → This paper addresses the problem of inferring unseen cross-modal image-to-image translations between multiple modalities. We assume that only some of the pairwise translations have been seen (i.e. trained) and infer the remaining unseen translations (where training pairs are not available). We propose mix and match […]

  • Generative Feature Replay For Class-Incremental Learning

    Xialei Liu, Chenshen Wu, Mikel Menta, Luis Herranz, Bogdan Raducanu, Andrew D. Bagdanov, Shangling Jui, Joost van de Weijer Read Full Paper → Humans are capable of learning new tasks without forgetting previous ones, while neural networks fail due to catastrophic forgetting between new and previously-learned tasks. We consider a class-incremental setting which means that the task-ID is unknown at inference time. […]

  • Semi-supervised Learning for Few-shot Image-to-Image Translation

    Yaxing Wang, Salman Khan, Abel Gonzalez-Garcia, Joost van de Weijer, Fahad Shahbaz Khan Read Full Paper → In the last few years, unpaired image-to-image translation has witnessed remarkable progress. Although the latest methods are able to generate realistic images, they crucially rely on a large number of labeled images. Recently, some methods have tackled the challenging setting of few-shot […]

  • Semantic Drift Compensation for Class-Incremental Learning

    Lu Yu, Bartłomiej Twardowski, Xialei Liu, Luis Herranz, Kai Wang, Yongmei Cheng, Shangling Jui, Joost van de Weijer Read Full Paper → Class-incremental learning of deep networks sequentially increases the number of classes to be classified. During training, the network has only access to data of one task at a time, where each task contains several classes. In this setting, networks suffer […]

  • Orderless Recurrent Models for Multi-label Classification

    Vacit Oguz Yazici, Abel Gonzalez-Garcia, Arnau Ramisa, Bartlomiej Twardowski, Joost van de Weijer Read Full Paper → Recurrent neural networks (RNN) are popular for many computer vision tasks, including multi-label classification. Since RNNs produce sequential outputs, labels need to be ordered for the multi-label classification task. Current approaches sort labels according to their frequency, typically ordering them in either […]

  • MineGAN: effective knowledge transfer from GANs to target domains with few images

    Yaxing Wang, Abel Gonzalez-Garcia, David Berga, Luis Herranz, Fahad Shahbaz Khan, Joost van de Weijer Read Full Paper → One of the attractive characteristics of deep neural networks is their ability to transfer knowledge obtained in one domain to other related domains. As a result, high-quality networks can be trained in domains with relatively little training data. This property has […]