Optimized Generic Feature Learning for Few-shot Classification across Domains

By Tonmoy Saikia, Thomas Brox, Cordelia Schmid. 2020

51 citations recorded by the source corpus.

To learn models or features that generalize across tasks and domains is one of the grand goals of machine learning. In this paper, we propose to use cross-domain, cross-task data as validation objective for hyper-parameter optimization (HPO) to improve on this goal. Given a rich enough search space, optimization of hyper-parameters learn features that maximize validation performance and, due to the objective, generalize across tasks and domains. We demonstrate the effectiveness of this strategy on few-shot image classification within and across domains. The learned features outperform all previous few-shot and meta-learning approaches.

Subfields: Artificial Intelligence, Computer Vision and Pattern Recognition, Meta & Continual Learning

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

1. Yoshua Bengio
2. Mohit Bansal
3. Quoc V. Le
4. Percy Liang
5. Ali Farhadi
6. Trevor Darrell
7. Sergey Levine
8. Ruslan Salakhutdinov

In Computer Vision and Pattern Recognition

1. Luc Van Gool
2. Dacheng Tao
3. Andrea Vedaldi
4. Trevor Darrell
5. Andrew Zisserman
6. Ming-Hsuan Yang
7. Xiaogang Wang
8. Jitendra Malik

In Meta & Continual Learning

1. Hakan Bilen
2. Xialei Liu
3. Wei-Hong Li
4. Ali Farhadi
5. Jasper Snoek
6. Mitchell Wortsman
7. Andrew Zisserman
8. Hugo Larochelle

Citations in the corpus

• Model soups: averaging weights of multiple fine-tuned models improves accuracy without increasing inference time (2022, methodology): “Finally, previous work has explored building ensembles of models produced by hyperparameter searches (Snoek et al., 2015; Mendoza et al., 2016; Saikia et al., 2020), including greedy selection strategies (Caruana et al.”
• CrossTransformers: spatially-aware few-shot transfer (2020, methodology): “Adding SimCLR episodes (+SimCLR Eps) and data augmentation inspired by BOHB [68] (+Aug) further improves results.”
• Improving Task Adaptation for Cross-domain Few-shot Learning (2021, methodology): “BOHB [38] proposes to use multi-domain data as validation objective for hyper-parameter optimization.”
• Hyperparameter Ensembles for Robustness and Uncertainty Quantification (2020, methodology): “, [19] and specifically for neural networks [53] as well as in computer vision [60] and genetics [35].”
• Learning a Universal Template for Few-shot Dataset Generalization (2021, background): “…> . work is shifting towards a more challenging instance of few-shot classification across datasets (Chen et al., 2019; Tseng et al., 2020; Triantafillou et al., 2020; Requeima et al., 2019; Bateni et al., 2020; Bronskill et al., 2020; Saikia et al., 2020; Dvornik et al., 2020; Liu et al., 2021).”
• A Universal Representation Transformer Layer for Few-Shot Image Classification (2020, methodology): “strategies [17] and more flexible meta-learning algorithms [16].”