Fast Adaptive Meta-Learning for Few-Shot Image Generation

Phaphuangwittayakul, Aniwat; Guo, Yi; Ying, Fangli

doi:10.1109/tmm.2021.3077729

Cited by 31 publications

(9 citation statements)

References 42 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…For one-and fewshot learning, a robust 'transfer of particle', introduced in 1901 by Woodworth [91], is also a desired mechanism because generalizing based on one or few datapoints to account for intra-class variability of thousands images is formidable. One-and Few-shot Learning (FSL) have been studied in computer vision in both shallow [12,2,11,45] and deep learning scenarios [32,84,73,13,78,107,24,23,57,69,80,56,25,94]. For brevity, we review only the deep learning techniques.…”

Section: A Learning From Few Samplesmentioning

confidence: 99%

“…LRPABN [24] uses an effective low-rank pairwise bilinear pooling operation to capture the nuanced differences between images. FAML [57] proposes a novel GAN-based few-shot image generation approach, which is capable of generating new realistic images for unseen target classes in the low-sample regime. Zhu et al [104] propose a novel global grouping metric to incorporate the global context, resulting in a per-channel modulation of local relation features.…”

Section: A Learning From Few Samplesmentioning

confidence: 99%

See 1 more Smart Citation

Multi-Level Second-Order Few-Shot Learning

Zhang

Koniusz

2023

IEEE Trans. Multimedia

View full text Add to dashboard Cite

We propose a Multi-level Second-order (MlSo) fewshot learning network for supervised or unsupervised few-shot image classification and few-shot action recognition. We leverage so-called power-normalized second-order base learner streams combined with features that express multiple levels of visual abstraction, and we use self-supervised discriminating mechanisms. As Second-order Pooling (SoP) is popular in image recognition, we employ its basic element-wise variant in our pipeline. The goal of multi-level feature design is to extract feature representations at different layer-wise levels of CNN, realizing several levels of visual abstraction to achieve robust few-shot learning. As SoP can handle convolutional feature maps of varying spatial sizes, we also introduce image inputs at multiple spatial scales into MlSo. To exploit the discriminative information from multi-level and multiscale features, we develop a Feature Matching (FM) module that reweights their respective branches. We also introduce a selfsupervised step, which is a discriminator of the spatial level and the scale of abstraction. Our pipeline is trained in an end-to-end manner. With a simple architecture, we demonstrate respectable results on standard datasets such as Omniglot, mini-ImageNet, tiered-ImageNet, Open MIC, fine-grained datasets such as CUB Birds, Stanford Dogs and Cars, and action recognition datasets such as HMDB51, UCF101, and mini-MIT.

show abstract

Section: A Learning From Few Samplesmentioning

confidence: 99%

Section: A Learning From Few Samplesmentioning

confidence: 99%

Multi-Level Second-Order Few-Shot Learning

Zhang

Koniusz

2023

IEEE Trans. Multimedia

View full text Add to dashboard Cite

show abstract

“…Meta learning has been applied to a range of research domains including various computer vision tasks, natural language processing and recently automatic speech recognition. In the computer vision area, meta learning has been exploited for the few-shot image classification task [12], object detection [13] and video generation [14]. In the natural language processing domain, meta learning has shown promising results in neural machine translation (NMT) for resource constraint languages [15].…”

Section: Meta Learningmentioning

confidence: 99%

Improved Meta Learning for Low Resource Speech Recognition

Singh

Wang

Feng

2022

ICASSP 2022 - 2022 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)

View full text Add to dashboard Cite

We propose a new meta learning based framework for low resource speech recognition that improves the previous model agnostic meta learning (MAML) approach. The MAML is a simple yet powerful meta learning approach. However, the MAML presents some core deficiencies such as training instabilities and slower convergence speed. To address these issues, we adopt multi-step loss (MSL). The MSL aims to calculate losses at every step of the inner loop of MAML and then combines them with a weighted importance vector. The importance vector ensures that the loss at the last step has more importance than the previous steps. Our empirical evaluation shows that MSL significantly improves the stability of the training procedure and it thus also improves the accuracy of the overall system. Our proposed system outperforms MAML based low resource ASR system on various languages in terms of character error rates and stable training behavior.

show abstract

“…The approach of few-shot learning is to train a network on the training set and fine-tune with the few data in the novel classes. Such scenarios can be applied to many tasks, such as video classification [18], image recognition [19], [20], image generation [21] or image semantic segmentation [1], [4], [22]. Few-shot semantic segmentation aims to perform pixel-level classification for novel classes in a query image conditioned on only a few annotated support images.…”

Section: Few-shot Segmentationmentioning

confidence: 99%

APANet: Adaptive Prototypes Alignment Network for Few-Shot Semantic Segmentation

Chen¹,

Gao²,

Lu³

et al. 2021

Preprint

View full text Add to dashboard Cite

Fast Adaptive Meta-Learning for Few-Shot Image Generation

Cited by 31 publications

References 42 publications

Multi-Level Second-Order Few-Shot Learning

Multi-Level Second-Order Few-Shot Learning

Improved Meta Learning for Low Resource Speech Recognition

APANet: Adaptive Prototypes Alignment Network for Few-Shot Semantic Segmentation

Contact Info

Product

Resources

About