One, few-shot learning

Last updated Apr 10, 2022 Edit Source

One-shot learning is an object categorization problem, found mostly in computer vision. Whereas most machine learning based object categorization algorithms require training on hundreds or thousands of samples/images and very large datasets, one-shot learning aims to learn information about object categories from one, or only a few, training samples/images.

• https://en.wikipedia.org/wiki/One-shot_learning
• https://medium.com/sap-machine-learning-research/deep-few-shot-learning-a1caa289f18
• https://sorenbouma.github.io/blog/oneshot/
• https://github.com/Goldesel23/Siamese-Networks-for-One-Shot-Learning
• Zero-Shot Visual Imitation
• #TALK Neural Networks - One Shot Learning

• #CODE LibFewShot
  • #PAPER LibFewShot: A Comprehensive Library for Few-shot Learning (Li 2021)
  • LibFewShot: A Comprehensive Library for Few-shot Learning (pytorch)

• #PAPER Siamese Neural Networks for One-shot Image Recognition (Koch 2015)
  • Few-Shot learning has seen great progress over the last years. A classic approach is based on metric learning using Siamese neural networks.
  • #CODE https://sorenbouma.github.io/blog/oneshot/
• #PAPER One-Shot Imitation Learning (Duan 2017)
• #PAPER One-shot texture segmentation (Ustyuzhaninov 2018)
  • We solve the task of one-shot texture segmentation in three steps. First, we compute embeddings of an input image and a reference patch; second, we search for the reference texture in the embedding space to produce a rough segmentation mask; and, finally, we employ a decoding network to produce the output segmentation.
• #PAPER One-shot instance segmentation (Michaelis 2018)
  • We tackle one-shot visual search by example for arbitrary object categories: Given an example image of a novel reference object, find and segment all object instances of the same category within a scene. To address this problem, we propose Siamese Mask R-CNN.
  • It extends Mask R-CNN by a Siamese backbone encoding both reference image and scene, allowing it to target detection and segmentation towards the reference category.
• #PAPER FIGR: Few-shot Image Generation with Reptile (Clouatre 2019)
• #PAPER Generalizing from a Few Examples: A Survey on Few-Shot Learning (Wang 2020)
• #PAPER ‘Less Than One’-Shot Learning: Learning N Classes From M < N Samples (Sucholutsky 2020)
  • https://www.technologyreview.com/2020/10/16/1010566/ai-machine-learning-with-tiny-data/
  • AI model able to accurately recognize more objects than the number of examples it was trained on
  • The trick, was to create images that blend multiple digits together and then feed them into an AI model with hybrid, or “soft,” labels

• #PAPER MetaGAN: An Adversarial Approach to Few-Shot Learning (Zhang 2018)
• #PAPER SinGAN: Learning a Generative Model from a Single Natural Image, SinGAN (Rott Shaham, ICCV 2019 Best Paper)
  • Paper explained
  • Paper explained
• #PAPER DAWSON: A Domain Adaptive Few Shot Generation Framework (Liang 2020)
• #PAPER LARGE: Latent-Based Regression through GAN Semantics (Nitzan 2021)
  • #CODE https://github.com/YotamNitzan/LARGE