Active One-shot Learning

By Mark P. Woodward, Chelsea Finn. 2017

135 citations recorded by the source corpus.

Recent advances in one-shot learning have produced models that can learn from a handful of labeled examples, for passive classification and regression tasks. This paper combines reinforcement learning with one-shot learning, allowing the model to decide, during classification, which examples are worth labeling. We introduce a classification task in which a stream of images are presented and, on each time step, a decision must be made to either predict a label or pay to receive the correct label. We present a recurrent neural network based action-value function, and demonstrate its ability to learn how and when to request labels. Through the choice of reward function, the model can achieve a higher prediction accuracy than a similar model on a purely supervised task, or trade prediction accuracy for fewer label requests.

Subfields: Artificial Intelligence

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

Citations in the corpus

Learning Algorithms for Active Learning (2017, methodology): “As in the recent one-shot learning work of Santoro et al. (2016) and Vinyals et al. (2016), and the active learning work of Woodward & Finn (2016), we evaluate our model on the Omniglot dataset.”
Generalizing from a Few Examples (2019, background): “Existing works mainly deal with computer vision applications, such as character recognition [15, 33, 34, 56, 61, 82, 99, 100, 105, 108, 118, 127, 134] and image classification [34, 61, 82, 93, 105, 108, 115, 118, 120, 127, 131, 132, 136].”
A Survey on Active Learning and Human-in-the-Loop Deep Learning for Medical Image Analysis (2019, methodology): “Woodward et al. (2017) propose a one-shot learning method that combines with RL to allow the model to decide, during inference, which examples are worth labelling.”
The Omniglot Challenge: A 3-Year Progress Report (2019, background): “…been contributed: Santoro et al. (2016) and Rae et al. (2016) studied sequential one-shot classiﬁcation with Omniglot where the stimuli arrive sequentially and agents must remember and generalize the category labels, and Woodward and Finn (2016) studied an active learning version of the same task.”
Deep Learning for Encrypted Traffic Classification: An Overview (2018, background): “ labeling them. In the ML community, active learning, where a model selects which data points should be labeled, has been studied for many years. In a recent study on classifying images of characters [15], a combination of reinforcement learning and LSTM is used to perform one of the two possible ac”
What Do I Annotate Next? An Empirical Study of Active Learning for Action Localization (2018, background): “Inspired by very recent ideas in active learning [1,25,40,70,74], our proposed active learning framework learns (or bootstrap) a function that selects samples for annotation based on knowledge extracted from a previous dataset.”
A Comprehensive Survey of Few-shot Learning: Evolution, Applications, Challenges, and Opportunities (2022, unknown): “One is to use generative models to caste prior knowledge [13], [14], [15], where bayesian programming learning [16] is the most representative framework [17] in this field.”
Variable-Shot Adaptation for Online Meta-Learning (2020, methodology): “This setting has been considered in black-box methods that receive datapoints and make predictions incrementally (Santoro et al., 2016; Woodward & Finn, 2017), but this approach is known to underperform the setting where the model is trained for a specific value of K (Mishra et al., 2017).”
Interactive Language Acquisition with One-shot Visual Concept Learning through a Conversational Game (2018, background): “Woodward and Finn (2016) present efforts toward active learning, using a vanilla recurrent neural network (RNN) without an external memory.”
Improving One-Shot Learning through Fusing Side Information (2017, background): “Woodward & Finn (2016) further extended MANN to learning to learn an active learner by using reinforcement learning.”
Towards a Continuous Knowledge Learning Engine for Chatbots (2018, background): “7Unlike existing RL-based interactive or active learning (Li et al., 2016b; Woodward and Finn, 2017), the user doesn’t provide feedback to guide the learning process of LiLi.”