An Overview of Deep Learning Architectures in Few-Shot Learning Domain

Abstract

Since 2012, Deep learning has revolutionized Artificial Intelligence and has achieved state-of-the-art outcomes in different domains, ranging from Image Classification to Speech Generation. Though it has many potentials, our current architectures come with the pre-requisite of large amounts of data. Few-Shot Learning (also known as one-shot learning) is a subfield of machine learning that aims to create such models that can learn the desired objective with less data, similar to how humans learn. In this paper, we have reviewed some of the well-known deep learning-based approaches towards few-shot learning. We have discussed the recent achievements, challenges, and possibilities of improvement of few-shot learning based deep learning architectures. Our aim for this paper is threefold: (i) Give a brief introduction to deep learning architectures for few-shot learning with pointers to core references. (ii) Indicate how deep learning has been applied to the low-data regime, from data preparation to model training. and, (iii) Provide a starting point for people interested in experimenting and perhaps contributing to the field of few-shot learning by pointing out some useful resources and open-source code. Our code is available at Github:https: //github.com/shruti-jadon/Hands-on-One-Shot-Learning

AdaptSum: Towards Low-Resource Domain Adaptation for Abstractive Summarization

“…Domain adaption for natural language processing and computer vision tasks is widely studied (Blitzer et al, 2007;Mansour et al, 2008;Daumé III, 2009;Sandu et al, 2010;Foster et al, 2010;Wang and Cardie, 2013;Sun et al, 2016;Liu et al, 2019bLiu et al, , 2020bGururangan et al, 2020;Winata et al, 2020;Jadon, 2020;Yin, 2020;Liu et al, 2020a,d). However, little has been done to investigate domain adaption for the abstractive summarization task.…”

Analysis of Vibrations and Currents for Broken Rotor Bar Detection in Three-phase Induction Motors

“…In this equation, Y is the label of a given pair (0 for negative pairs and 1 for positive pairs), D w is a similarity index describing the similarity between the embeddigs of the samples present in the pair and m is parameter known as margin. This function is consisted of two terms; the first term is supposed to represent observations of similar classes as closely possible, while the second term is responsible to increase the dissimilarity of the observations from different classes up to the highest extent (Jadon, 2020).…”

Noise-Robust Representation for Fault Identification with Limited Data via Data Augmentation

“…Therefore, in the constructed embedding by supervised contrastive learning, samples with the same label are placed near each other and those with different labels are placed far apart from each other. This approach is an example of few-shot learning methods (Jadon, 2020). Unsupervised contrastive learning incorporates data augmentation.…”

“…According to the contrastive loss equation, the first term is supposed to represent similar observations as closely as possible, while the second term is supposed to increase the distance between dissimilar observations. (Jadon, 2020).…”