Task-Agnostic Meta-Learning for Few-shot Learning

Jamal, Muhammad Abdullah; Qi, Guo-Jun; Shah, Mubarak

doi:10.48550/arxiv.1805.07722

Cited by 5 publications

(7 citation statements)

References 0 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In Table 1 we also include the results of our own implementation of MAML, which reproduces all results except the 20-way 1-shot Omniglot case. Difficulty in replicating the specific result has also been noted before in Jamal et al (2018). We base our Table 1: MAML++ Omniglot 20-way Few-Shot Results: Our reproduction of MAML appears to be replicating all the results except the 20-way 1-shot results.…”

Section: Resultsmentioning

confidence: 61%

“…We base our Table 1: MAML++ Omniglot 20-way Few-Shot Results: Our reproduction of MAML appears to be replicating all the results except the 20-way 1-shot results. Other authors have come across this problem as well Jamal et al (2018). We report our own base-lines to provide better relative intuition on how each method impacted the test accuracy of the model.…”

Section: Resultsmentioning

confidence: 99%

See 1 more Smart Citation

How to train your MAML

Antoniou¹,

Edwards²,

Storkey³

2018

Preprint

128

View full text Add to dashboard Cite

The field of few-shot learning has recently seen substantial advancements. Most of these advancements came from casting few-shot learning as a meta-learning problem. Model Agnostic Meta Learning or MAML is currently one of the best approaches for few-shot learning via meta-learning. MAML is simple, elegant and very powerful, however, it has a variety of issues, such as being very sensitive to neural network architectures, often leading to instability during training, requiring arduous hyperparameter searches to stabilize training and achieve high generalization and being very computationally expensive at both training and inference times. In this paper, we propose various modifications to MAML that not only stabilize the system, but also substantially improve the generalization performance, convergence speed and computational overhead of MAML, which we call MAML++.

show abstract

Section: Resultsmentioning

confidence: 61%

Section: Resultsmentioning

confidence: 99%

How to train your MAML

Antoniou¹,

Edwards²,

Storkey³

2018

Preprint

128

View full text Add to dashboard Cite

show abstract

“…Furthermore, extended (Finn et al, 2017) by introducing Bayesian mechanisms for fast adaptation and meta-update, quickly obtaining an approximate posterior of a given unseen task, as well a probabilistic framework was developed by . To avoid a biased meta-learner like (Finn et al, 2017), Jamal et al (2018) proposed a task-agnostic meta-learning (TAML) algorithms to train a meta-learner unbiased towards a variety of tasks before its initial model was adapted to unseen tasks. To deal with the long-tail distribution in big data, Wang et al (2017c) introduced a meta-network that learned to progressively transfer meta-knowledge from the head to the tail classes, where meta-knowledge was encoded with a meta-network trained to predict many-shot model parameters from few-shot model parameters.…”

Section: Approach 4: Meta Learningmentioning

confidence: 99%

Small Sample Learning in Big Data Era

Shu,

Xu,

Meng

2018

Preprint

View full text Add to dashboard Cite

As a promising area in artificial intelligence, a new learning paradigm, called Small Sample Learning (SSL), has been attracting prominent research attention in the recent years. In this paper, we aim to present a survey to comprehensively introduce the current techniques proposed on this topic. Specifically, current SSL techniques can be mainly divided into two categories. The first category of SSL approaches can be called "concept learning", which emphasizes learning new concepts from only few related observations. The purpose is mainly to simulate human learning behaviors like recognition, generation, imagination, synthesis and analysis. The second category is called "experience learning", which usually co-exists with the large sample learning manner of conventional machine learning. This category mainly focuses on learning with insufficient samples, and can also be called small data learning in some literatures. More extensive surveys on both categories of SSL techniques are introduced and some neuroscience evidences are provided to clarify the rationality of the entire SSL regime, and the relationship with human learning process. Some discussions on the main challenges and possible future research directions along this line are also presented.

show abstract

“…A popular meta-learning strategy consists of finding model initialisations that allow fast adaptation to new, previously unseen tasks [17]. The strategy has since been widely adopted for classification tasks [29] and several recently proposed extensions report increases to efficiency [33] and performance [31,2]. While a natural separation of few-shot tasks exists for image classification problems, in contrast, problems framed as a regression (e.g.…”

Section: Related Workmentioning

confidence: 99%

Formulating Camera-Adaptive Color Constancy as a Few-shot Meta-Learning Problem

McDonagh,

Parisot,

Zhou

et al. 2018

Preprint

View full text Add to dashboard Cite

Digital camera pipelines employ color constancy methods to estimate an unknown scene illuminant, in order to reilluminate images as if they were acquired under an achromatic light source. Fully-supervised learning approaches exhibit state-of-the-art estimation accuracy with cameraspecific labelled training imagery. Resulting models typically suffer from domain gaps and fail to generalise across imaging devices. In this work, we propose a new approach that affords fast adaptation to previously unseen cameras, and robustness to changes in capture device by leveraging annotated samples across different cameras and datasets. We present a general approach that utilizes the concept of color temperature to frame color constancy as a set of distinct, homogeneous few-shot regression tasks, each associated with an intuitive physical meaning. We integrate this novel formulation within a meta-learning framework, enabling fast generalisation to previously unseen cameras using only handfuls of camera specific training samples. Consequently, the time spent for data collection and annotation substantially diminishes in practice whenever a new sensor is used. To quantify this gain, we evaluate our pipeline on three publicly available datasets comprising 12 different cameras and diverse scene content. Our approach delivers competitive results both qualitatively and quantitatively while requiring a small fraction of the camera-specific samples compared to standard approaches.

show abstract

Task-Agnostic Meta-Learning for Few-shot Learning

Cited by 5 publications

References 0 publications

How to train your MAML

How to train your MAML

Small Sample Learning in Big Data Era

Formulating Camera-Adaptive Color Constancy as a Few-shot Meta-Learning Problem

Contact Info

Product

Resources

About