Filters








11,897 Hits in 5.8 sec

Improving Few-Shot Learning with Auxiliary Self-Supervised Pretext Tasks [article]

Nathaniel Simard, Guillaume Lagrange
2021 arXiv   pre-print
Recent work on few-shot learning showed that quality of learned representations plays an important role in few-shot classification performance.  ...  In this work, we exploit the complementarity of both paradigms via a multi-task framework where we leverage recent self-supervised methods as auxiliary tasks.  ...  Conclusion Based on the simple baseline of Tian et al. (2020a) , we have proposed a multi-task framework with self-supervised auxiliary tasks to improve few-shot image classification.  ... 
arXiv:2101.09825v1 fatcat:gkmrz3z4dja55j5uycsrqcmgcm

Self-Supervised Meta-Learning for Few-Shot Natural Language Classification Tasks [article]

Trapit Bansal, Rishikesh Jha, Tsendsuren Munkhdalai, Andrew McCallum
2020 arXiv   pre-print
Furthermore, we show how the self-supervised tasks can be combined with supervised tasks for meta-learning, providing substantial accuracy gains over previous supervised meta-learning.  ...  This paper proposes a self-supervised approach to generate a large, rich, meta-learning task distribution from unlabeled text.  ...  (2019) demonstrated that better feature learning from supervised tasks helps few-shot learning. Thus, we also evaluate multi-task learning and multi-task meta-learning for few-shot generalization.  ... 
arXiv:2009.08445v2 fatcat:klscagonaveaxo67swdr56pyry

Multi-Pretext Attention Network for Few-shot Learning with Self-supervision [article]

Hainan Li, Renshuai Tao, Jun Li, Haotong Qin, Yifu Ding, Shuo Wang, Xianglong Liu
2021 arXiv   pre-print
Few-shot learning is an interesting and challenging study, which enables machines to learn from few samples like humans.  ...  Self-supervised learning is emerged as an efficient method to utilize unlabeled data.  ...  Inspired by the similarity of few-shot and self-supervised learning, some works [6, 7] have weaved self-supervision into the training process of few-shot learning.  ... 
arXiv:2103.05985v1 fatcat:wgadzl75gzeobcdyvqypyzxt2e

Pareto Self-Supervised Training for Few-Shot Learning [article]

Zhengyu Chen, Jixie Ge, Heshen Zhan, Siteng Huang, Donglin Wang
2021 arXiv   pre-print
While few-shot learning (FSL) aims for rapid generalization to new concepts with little supervision, self-supervised learning (SSL) constructs supervisory signals directly computed from unlabeled data.  ...  Exploiting the complementarity of these two manners, few-shot auxiliary learning has recently drawn much attention to deal with few labeled data.  ...  Boosting few- shot visual learning with self-supervision.  ... 
arXiv:2104.07841v2 fatcat:gywa2f3ikvf6fluxbdhjjjcoea

Visual Representation Learning with Self-Supervised Attention for Low-Label High-data Regime [article]

Prarthana Bhattacharyya, Chenge Li, Xiaonan Zhao, István Fehérvári, Jason Sun
2022 arXiv   pre-print
In this paper, we are the first to question if self-supervised vision transformers (SSL-ViTs) can be adapted to two important computer vision tasks in the low-label, high-data regime: few-shot image classification  ...  Our self-supervised attention representations outperforms the state-of-the-art on several public benchmarks for both tasks, namely miniImageNet and CUB200 for few-shot image classification by up-to 6 CUB200  ...  Few-Shot Image Classification with Self-Supervised Feature Embeddings: For our few-shot image-classification framework, we follow the distribution calibration [9] methodology proposed as an alternative  ... 
arXiv:2201.08951v2 fatcat:6ulgcp5ornh53hyt6ml6uwkada

CSN: Component-Supervised Network for Few-Shot Classification [article]

Shuai Shao, Baodi Liu, Lei Xing, Lifei Zhao, Yanjiang Wang, Weifeng Liu, Yicong Zhou
2022 arXiv   pre-print
The few-shot classification (FSC) task has been a hot research topic in recent years. It aims to address the classification problem with insufficient labeled data on a cross-category basis.  ...  Starting from the root cause of this problem, this paper presents a new scheme, Component-Supervised Network (CSN), to improve the performance of FSC.  ...  recently proposed semi-supervised few-shot classification methods.  ... 
arXiv:2203.07738v1 fatcat:l74sp5dvpfazhl3ztk4jxn7ulu

Rethinking Few-Shot Image Classification: a Good Embedding Is All You Need? [article]

Yonglong Tian, Yue Wang, Dilip Krishnan, Joshua B. Tenenbaum, Phillip Isola
2020 arXiv   pre-print
state-of-the-art few-shot learning methods.  ...  We believe that our findings motivate a rethinking of few-shot image classification benchmarks and the associated role of meta-learning algorithms.  ...  classification or self-supervised learning, on this combined dataset.  ... 
arXiv:2003.11539v2 fatcat:y3d2r3kpdjgsjnedlqbfse4774

When Does Self-supervision Improve Few-shot Learning? [article]

Jong-Chyi Su, Subhransu Maji, Bharath Hariharan
2020 arXiv   pre-print
We investigate the role of self-supervised learning (SSL) in the context of few-shot learning.  ...  We find that SSL reduces the relative error rate of few-shot meta-learners by 4%-27%, even when the datasets are small and only utilizing images within the datasets.  ...  Few-shot learning as an evaluation for self-supervised tasks The fewshot classification task provides a way of evaluating the effectiveness of selfsupervised tasks.  ... 
arXiv:1910.03560v2 fatcat:wt4oebe5ejcptdgxfhpghbldty

Representation Based Meta-Learning for Few-Shot Spoken Intent Recognition

Ashish Mittal, Samarth Bharadwaj, Shreya Khare, Saneem Chemmengath, Karthik Sankaranarayanan, Brian Kingsbury
2020 Interspeech 2020  
This paper presents a few-shot spoken intent classification approach with task-agnostic representations via meta-learning paradigm.  ...  The performance is comparable to traditionally supervised classification models with abundant training samples.  ...  We also hypothesize a weighted combination of the reconstruction loss from the self-supervision (with a controlling parameter α) together with the cross-entropy loss from the meta-learning few-shot classification  ... 
doi:10.21437/interspeech.2020-3208 dblp:conf/interspeech/MittalBKCSK20 fatcat:2zerbq2eh5a6rbco7a2jjqrlqu

Label-Efficient Learning on Point Clouds using Approximate Convex Decompositions [article]

Matheus Gadelha, Aruni RoyChowdhury, Gopal Sharma, Evangelos Kalogerakis, Liangliang Cao, Erik Learned-Miller, Rui Wang, Subhransu Maji
2020 arXiv   pre-print
We show that using ACD to approximate ground truth segmentation provides excellent self-supervision for learning 3D point cloud representations that are highly effective on downstream tasks.  ...  We report improvements over the state-of-the-art for unsupervised representation learning on the ModelNet40 shape classification dataset and significant gains in few-shot part segmentation on the ShapeNetPart  ...  ., the few-shot classification setting, including self-supervised losses along with the usual supervised training is shown to be beneficial [59] .  ... 
arXiv:2003.13834v2 fatcat:6wbuhbjsmfcjrghskkspxbf3l4

Self-Supervised Learning For Few-Shot Image Classification [article]

Da Chen, Yuefeng Chen, Yuhong Li, Feng Mao, Yuan He, Hui Xue
2021 arXiv   pre-print
In this paper, we proposed to train a more generalized embedding network with self-supervised learning (SSL) which can provide robust representation for downstream tasks by learning from the data itself  ...  Few-shot image classification aims to classify unseen classes with limited labelled samples.  ...  A popular strategy for few-shot learning is through metalearning (also called learning-to-learn) with multi-auxiliary tasks [11, 12, 13, 3] .  ... 
arXiv:1911.06045v3 fatcat:ac5q4scuofdynk2rkqjce4kxlm

An Efficient Method for the Classification of Croplands in Scarce-Label Regions [article]

Houtan Ghaffari
2021 arXiv   pre-print
We introduce three self-supervised tasks for cropland classification.  ...  We will show how to leverage their potential for cropland classification using self-supervised tasks.  ...  [33] , proposed to train the self-supervised tasks jointly with the main one, just like the way we used for few-shot learning.  ... 
arXiv:2103.09588v1 fatcat:erzaggmwgrabxfcwrorr4zdtdi

Self Supervision to Distillation for Long-Tailed Visual Recognition [article]

Tianhao Li, Limin Wang, Gangshan Wu
2021 arXiv   pre-print
Specifically, we propose a conceptually simple yet particularly effective multi-stage training scheme, termed as Self Supervised to Distillation (SSD). This scheme is composed of two parts.  ...  Second, we present a new distillation label generation module guided by self-supervision.  ...  Feature learning enhanced by self-supervision In phase-I of the feature learning stage, we choose to train the backbone network using a standard supervised task and a self-supervised task in a multi-task  ... 
arXiv:2109.04075v1 fatcat:5gi4s55pcfaixbcrwfvss5erym

SAFFNet: Self-Attention-Based Feature Fusion Network for Remote Sensing Few-Shot Scene Classification

Joseph Kim, Mingmin Chi
2021 Remote Sensing  
Here, the feature weighting value can be fine-tuned by the support set in the few-shot learning task.  ...  In this paper, a multi-scale feature fusion network for few-shot remote sensing scene classification is proposed by integrating a novel self-attention feature selection module, denoted as SAFFNet.  ...  Comparison of traditional supervised, zero-shot and few-shot learning for classification tasks.  ... 
doi:10.3390/rs13132532 fatcat:fonmvysiczct5kr7wwl3xe2xdm

Learning to Few-Shot Learn Across Diverse Natural Language Classification Tasks [article]

Trapit Bansal, Rishikesh Jha, Andrew McCallum
2020 arXiv   pre-print
parameters for few-shot learning than self-supervised pre-training or multi-task training, outperforming many strong baselines, for example, yielding 14.5% average relative gain in accuracy on unseen tasks  ...  We consider this problem of learning to generalize to new tasks with few examples as a meta-learning problem.  ...  inference, sentiment classification, and various other text classification tasks; (4) we study how metalearning, multi-task learning and fine-tuning perform for few-shot learning of completely new tasks  ... 
arXiv:1911.03863v3 fatcat:7bppnqaqirfy7a3b2lekqxzp7i
« Previous Showing results 1 — 15 out of 11,897 results