, Prototype Matching Nets
Weight Gen 45 ,
Weight Gen 46 ,
, with and without priors) fot the ImageNet based few-shot benchmark proposed in [4] (for more details about the evaluation metrics we refer to [26]). For each novel category we use N = 1, 2, 5, 10 or 20 training examples. Methods with " w/ H " use mechanisms that hallucinate extra training examples for the novel categories, The second rows in our entries report the 95% confidence intervals
Learning to learn by gradient descent by gradient descent, Advances in Neural Information Processing Systems, pp.3981-3989, 2016. ,
Cosine normalization: Using cosine similarity instead of dot product in neural networks, 2017. ,
Model-agnostic metalearning for fast adaptation of deep networks. arXiv preprint, p.7, 2017. ,
Low-shot visual recognition by shrinking and hallucinating features. arXiv preprint arXiv:1606, 2008. ,
Deep Residual Learning for Image Recognition, 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), pp.770-778, 2008. ,
DOI : 10.1109/CVPR.2016.90
Long Short-Term Memory, Neural Computation, vol.4, issue.8, pp.1735-1780, 1997. ,
DOI : 10.1016/0893-6080(88)90007-X
Deep Metric Learning Using Triplet Network, International Workshop on Similarity-Based Pattern Recognition, pp.84-92, 2015. ,
DOI : 10.1145/1553374.1553469
URL : http://arxiv.org/pdf/1412.6622
Siamese neural networks for one-shot image recognition, ICML Deep Learning Workshop, 2015. ,
ImageNet classification with deep convolutional neural networks, Advances in neural information processing systems, pp.1097-1105, 2012. ,
DOI : 10.1162/neco.2009.10-08-881
Gradientbased learning applied to document recognition, Proceedings of the IEEE, pp.2278-2324, 1998. ,
Metric Learning for Large Scale Image Classification: Generalizing to New Classes at Near-Zero Cost, Computer Vision?ECCV 2012, pp.488-501, 2012. ,
DOI : 10.1007/978-3-642-33709-3_35
URL : https://hal.archives-ouvertes.fr/hal-00722313
Meta-learning with temporal convolutions. arXiv preprint, p.7, 2017. ,
Meta networks. arXiv preprint, 2017. ,
Rectified linear units improve restricted boltzmann machines, Proceedings of the 27th international conference on machine learning (ICML-10), pp.807-814, 2010. ,
Learning with imprinted weights. arXiv preprint, 2017. ,
Optimization as a model for fewshot learning, 2007. ,
iCaRL: Incremental Classifier and Representation Learning, 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), pp.2001-2010, 2017. ,
DOI : 10.1109/CVPR.2017.587
URL : http://arxiv.org/pdf/1611.07725
ImageNet Large Scale Visual Recognition Challenge, International Journal of Computer Vision, vol.1010, issue.1, pp.211-252, 2015. ,
DOI : 10.1007/978-3-642-15555-0_11
URL : http://dspace.mit.edu/bitstream/1721.1/104944/1/11263_2015_Article_816.pdf
One-shot learning with memory-augmented neural networks, 2016. ,
Shifting inductive bias with success-story algorithm, adaptive levin search, and incremental self-improvement, Machine Learning, pp.105-130, 1997. ,
Very deep convolutional networks for large-scale image recognition. arXiv preprint, 2014. ,
Prototypical networks for few-shot learning. arXiv preprint, 2008. ,
Going deeper with convolutions, 2015 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), pp.1-9, 2015. ,
DOI : 10.1109/CVPR.2015.7298594
URL : http://arxiv.org/pdf/1409.4842
Lifelong learning algorithms. Learning to learn, pp.181-209, 1998. ,
Matching networks for one shot learning, Advances in Neural Information Processing Systems, pp.3630-3638, 2008. ,
Low-shot learning from imaginary data, 2018. ,
Learning deep features for scene recognition using places database ,
DOI : 10.1109/tpami.2017.2723009
, Advances in Neural Information Processing Systems 27, pp.487-495, 2014.