Few-shot learning

65 results found

Sort

ICLR: The AI conference that helped redefine the field

Larry Hardesty

April 24, 2020

Amazon’s Stefano Soatto on how learning representations came to dominate machine learning.

Machine learning
Learning computational tasks from single examples

Pablo Garcia Moreno, Andreas Damianou

April 09, 2020

New “meta-learning” approach improves on the state of the art in “one-shot” learning.

Machine learning
Empirical Bayes transductive meta-learning with synthetic gradients

Xu Hu, Pablo Garcia Moreno, Xi Shen, Yang Xiao, Guillaume Obozinski, Neil Lawrence, Andreas Damianou

ICLR 2020

2020

We propose a meta-learning approach that learns from multiple tasks in a transductive setting, by leveraging the unlabeled query set in addition to the support set to generate a more powerful model for each task. To develop our framework, we revisit the empirical Bayes formulation for multi-task learning. The evidence lower bound of the marginal log-likelihood of empirical Bayes decomposes as a sum of local

Related: Learning computational tasks from single examples

Machine learning
A baseline for few-shot image classiﬁcation

Guneet Singh Dhillon, Pratik Chaudhari, Avinash Ravichandran, Stefano Soatto

ICLR 2020

2020

Fine-tuning a deep network trained with the standard cross-entropy loss is a strong baseline for few-shot learning. When ﬁne-tuned transductively, this outperforms the current state-of-the-art on standard datasets such as Mini-Imagenet, Tiered- Imagenet, CIFAR-FS and FC-100 with the same hyper-parameters. The simplicity of this approach enables us to demonstrate the ﬁrst few-shot learning results on the

Machine learning
Few-Shot Learning with Embedded Class Models and Shot-Free Meta Training

Avinash Ravichandran, Rahul Bhotika, Stefano Soatto

ICCV 2019

2019

We propose a method for learning embeddings for few-shot learning that is suitable for use with any number of shots (shot-free). Rather than ﬁxing the class prototypes to be the Euclidean average of sample embeddings, we allow them to live in a higher-dimensional space (embedded class models) and learn the prototypes along with the model parameters. The class representation function is deﬁned implicitly

Computer vision

...

Few-shot learning

Work with us