Abstract
Though machine learning algorithms have achieved great performance when adequate amounts of labeled data is available, there has been growing interest in reducing the volume of data required. While humans tend to be highly effective in this context, it remains a challenge for machine learning approaches. The goal of our work is to develop a visual learning based few-shot system that achieves good performance on novel few shot classes (with less than 5 samples each for training) and does not degrade the performance on the pre-trained large scale base classes and has a fast inference with little or zero training for adding new classes to the existing model. In this paper, we propose a novel, computationally efficient, yet effective framework called Param-Net, which is a multi-layer transformation function to convert the activations of a particular class to its corresponding parameters. Param-Net is pre-trained on large-scale base classes, and at inference time it adapts to novel few shot classes with just a single forward pass and zero-training, as the network is category-agnostic. Two publicly available datasets: MiniImageNet and Pascal-VOC were used for evaluation and benchmarking. Extensive comparison with related works indicate that, Param-Net outperforms the current state-of-the-art on 1-shot and 5-shot object recognition tasks in terms of accuracy as well as faster convergence (zero training). We also propose to fine-tune Param-Net with base classes as well as few-shot classes to significantly improve the accuracy (by more than 10% over zero-training approach), at the cost of slightly slower convergence (138 s of training on a Tesla K80 GPU for addition of a set of novel classes).
Supported by HCL Technologies Ltd.
This is a preview of subscription content, log in via an institution to check access.
Access this chapter
Tax calculation will be finalised at checkout
Purchases are for personal use only
Similar content being viewed by others
References
Sun, Q., Liu, Y., Chua, T.S., Schiele, B.: Meta-transfer learning for few-shot learning. In: Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, pp. 403–412 (2019)
Jamal, M.A., Qi, G.J.: Task agnostic meta-learning for few-shot learning. In: Proceedings of IEEE Conference on Computer Vision and Pattern Recognition (2019)
Lifchitz, Y., Avrithis, Y., Picard, S., Bursuc, A.: Dense classification and implanting for few-shot learning. In: Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, pp. 9258–9267 (2019)
Li, W., Wang, L., Xu, J., Huo, J., Gao, Y., Luo, J.: Revisiting local descriptor based image-to-class measure for few-shot learning. In: Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (2019)
Wertheimer, D., Hariharan, B.: Few-shot learning with localization in realistic settings. In: Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, pp. 6558–6567 (2019)
Gidaris, S., Komodakis, N.: Generating Classification Weights with GNN Denoising Autoencoders for Few-Shot Learning. arXiv preprint arXiv:1905.01102 (2019)
Kim, J., Kim, T., Kim, S., Yoo, C.D.: Edge-labeling graph neural network for few-shot learning. In: Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, pp. 11–20 (2019)
Li, H., Eigen, D., Dodge, S., Zeiler, M., Wang, X.: Finding task-relevant features for few-shot learning by category traversal. In: Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (2019)
Chu, W.H., Li, Y.J., Chang, J.C., Wang, Y.C.F.: Spot and learn: a maximum-entropy patch sampler for few-shot image classification. In: Proceedings of Conference on Computer Vision and Pattern Recognition (2019)
Zhang, H., Zhang, J., Koniusz, P.: Few-shot learning via saliency-guided hallucination of samples. In: Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, pp. 2770–2779 (2019)
Chen, Z., Fu, Y., Wang, Y.X., Ma, L., Liu, W., Hebert, M.: Image deformation meta-networks for one-shot learning. In: Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (2019)
Ravi, S., Larochelle, H.: Optimization as a model for few-shot learning (2016)
Vinyals, O., Blundell, C., Lillicrap, T., Wierstra, D.: Matching networks for one shot learning. In: Advances in Neural Information Processing Systems (2016)
Finn, C., Abbeel, P., Levine, S.: Model-agnostic meta-learning for fast adaptation of deep networks. In: Proceedings of the 34th International Conference on Machine Learning, vol. 70, pp. 1126–1135, August 2017
Snell, J., Swersky, K., Zemel, R.: Prototypical networks for few-shot learning. In: Advances in Neural Information Processing Systems, pp. 4077–4087 (2017)
Sung, F., Yang, Y., Zhang, L., Xiang, T., Torr, P.H., Hospedales, T.M.: Learning to compare: relation network for few-shot learning. In: Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (2018)
Mishra, N., Rohaninejad, M., Chen, X., Abbeel, P.: A simple neural attentive meta-learner. arXiv pre-print arXiv:1707.03141 (2017)
Franceschi, L., Frasconi, P., Salzo, S., Grazzi, R., Pontil, M.: Bilevel programming for hyperparameter optimization and meta-learning. arXiv preprint arXiv:1806.04910 (2018)
Zhang, R., Che, T., Ghahramani, Z., Bengio, Y., Song, Y.: MetaGAN: an adversarial approach to few-shot learning. In: Advances in Neural Information Processing Systems, pp. 2365–2374 (2018)
Munkhdalai, T., Yuan, X., Mehri, S., Trischler, A.: Rapid adaptation with conditionally shifted neurons. arXiv preprint arXiv:1712.09926 (2017)
Li, Z., Zhou, F., Chen, F., Li, H.: Meta-SGD: learning to learn quickly for few-shot learning. arXiv pre-print arXiv:1707.09835 (2017)
Gidaris, S., Komodakis, N.: Dynamic few-shot visual learning without forgetting. In: Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, pp. 4367–4375 (2018)
Garcia, V., Bruna, J.: Few-shot learning with graph neural networks. arXiv preprint arXiv:1711.04043 (2017)
Cai, Q., Pan, Y., Yao, T., Yan, C., Mei, T.: Memory matching networks for one-shot image recognition. In: Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, pp. 4080–4088 (2018)
Zhang, H., Koniusz, P.: Power normalizing second-order similarity network for few-shot learning. In: 2019 IEEE Winter Conference on Applications of Computer Vision (WACV), pp. 1185–1193. IEEE, January 2019
Nichol, A., Achiam, J., Schulman, J.: On first-order meta-learning algorithms. arXiv preprint arXiv:1803.02999 (2018)
Liu, Y., et al.: Learning to propagate labels: transductive propagation network for few-shot learning. arXiv preprint arXiv:1805.10002 (2018)
Jiang, X., Havaei, M., Varno, F., Chartrand, G., Chapados, N., Matwin, S.: Learning to learn with conditional class dependencies (2018)
Allen, K.R., Shin, H., Shelhamer, E., Tenenbaum, J.B.: Variadic meta-learning by Bayesian nonparametric deep embedding (2018)
Koch, G., Zemel, R., Salakhutdinov, R.: Siamese neural networks for one-shot image recognition. In: ICML Deep Learning Workshop, vol. 2, July 2015
Author information
Authors and Affiliations
Corresponding author
Editor information
Editors and Affiliations
Rights and permissions
Copyright information
© 2020 Springer Nature Switzerland AG
About this paper
Cite this paper
Kumar, N.S., Phirke, M.R., Jayapal, A., Thangam, V. (2020). Dynamic Visual Few-Shot Learning Through Parameter Prediction Network. In: Kotsireas, I., Pardalos, P. (eds) Learning and Intelligent Optimization. LION 2020. Lecture Notes in Computer Science(), vol 12096. Springer, Cham. https://doi.org/10.1007/978-3-030-53552-0_24
Download citation
DOI: https://doi.org/10.1007/978-3-030-53552-0_24
Published:
Publisher Name: Springer, Cham
Print ISBN: 978-3-030-53551-3
Online ISBN: 978-3-030-53552-0
eBook Packages: Mathematics and StatisticsMathematics and Statistics (R0)