Multi-Level Second-Order Few-Shot Learning (2022-01-15T00:00:00.000000Z)

TL;DR

A Multi-level Second-order (MlSo) few-shots learning network for supervised or unsupervised few-shot image classification and few- shot action recognition is proposed, leveraging so-called power-normalized second-order base learner streams combined with features that express multiple levels of visual abstraction, and self-supervised discriminating mechanisms are used.

Abstract

We propose a Multi-level Second-order (MlSo) few-shot learning network for supervised or unsupervised few-shot image classification and few-shot action recognition. We leverage so-called power-normalized second-order base learner streams combined with features that express multiple levels of visual abstraction, and we use self-supervised discriminating mechanisms. As Second-order Pooling (SoP) is popular in image recognition, we employ its basic element-wise variant in our pipeline. The goal of multi-level feature design is to extract feature representations at different layer-wise levels of CNN, realizing several levels of visual abstraction to achieve robust few-shot learning. As SoP can handle convolutional feature maps of varying spatial sizes, we also introduce image inputs at multiple spatial scales into MlSo. To exploit the discriminative information from multi-level and multi-scale features, we develop a Feature Matching (FM) module that reweights their respective branches. We also introduce a self-supervised step, which is a discriminator of the spatial level and the scale of abstraction. Our pipeline is trained in an end-to-end manner. With a simple architecture, we demonstrate respectable results on standard datasets such as Omniglot, mini–ImageNet, tiered–ImageNet, Open MIC, fine-grained datasets such as CUB Birds, Stanford Dogs and Cars, and action recognition datasets such as HMDB51, UCF101, and mini–MIT.

Authors

Hongdong Li

7 papers

Hongguang Zhang

3 papers

Piotr Koniusz

1 papers

TL;DR

Abstract

Authors

References117 items

ScaleNet: An Unsupervised Representation Learning Method for Limited Information

3D Skeleton-based Few-shot Action Recognition with JEANIE is not so Naïve

Fast Adaptive Meta-Learning for Few-Shot Image Generation

Tensor Representations for Action Recognition

Power Normalizations in Fine-Grained Image, Few-Shot Image and Graph Classification

Beyond Covariance: SICE and Kernel Based Visual Feature Representation

Multi-attention Meta Learning for Few-shot Fine-grained Image Recognition

Domain Adaptation

Host–Parasite: Graph LSTM-in-LSTM for Group Activity Recognition

DeepEMD: Few-Shot Image Classification With Differentiable Earth Mover’s Distance and Structured Classifiers

A Simple Framework for Contrastive Learning of Visual Representations

Relation Embedding for Personalised Translation-Based POI Recommendation

Received By

Rethinking Class Relations: Absolute-relative Supervised and Unsupervised Few-shot Learning

Few-Shot Action Recognition with Permutation-Invariant Attention

FLAT: Few-Shot Learning via Autoencoding Transformation Regularizers

Momentum Contrast for Unsupervised Visual Representation Learning

SimpleShot: Revisiting Nearest-Neighbor Classification for Few-Shot Learning

When Does Self-supervision Improve Few-shot Learning?

Few-Shot Image Recognition With Knowledge Transfer

Variational Few-Shot Learning

Low-Rank Pairwise Alignment Bilinear Network For Few-Shot Fine-Grained Image Classification

Coherence Constrained Graph LSTM for Group Activity Recognition

Boosting Few-Shot Visual Learning With Self-Supervision

Edge-Labeling Graph Neural Network for Few-Shot Learning

Generating Classification Weights With GNN Denoising Autoencoders for Few-Shot Learning

Meta-Learning With Differentiable Convex Optimization

Few-Shot Learning via Saliency-Guided Hallucination of Samples

Fisher-Bures Adversary Graph Convolutional Networks

Simplifying Graph Convolutional Networks

Unsupervised Meta-Learning for Few-Shot Image Classification

Power Normalizing Second-Order Similarity Network for Few-Shot Learning

Hierarchical Long Short-Term Concurrent Memory for Human Interaction Recognition

How to train your MAML

Incremental Few-Shot Learning with Attention Attractor Networks

Fast Context Adaptation via Meta-Learning

Predict then Propagate: Graph Neural Networks meet Personalized PageRank

Unsupervised Learning via Meta-Learning

Compound Memory Networks for Few-Shot Video Classification

Museum Exhibit Identification Challenge for the Supervised Domain Adaptation and Beyond

Neural Graph Matching Networks for Fewshot 3D Action Recognition

Second-order Democratic Aggregation

Meta-Learning with Latent Embedding Optimization

Dense Dilated Network for Few Shot Action Recognition

A Deeper Look at Power Normalizations

Learning to Propagate Labels: Transductive Propagation Network for Few-Shot Learning

TADAM: Task dependent adaptive metric for improved few-shot learning

Task Agnostic Meta-Learning for Few-Shot Learning

Dynamic Few-Shot Visual Learning Without Forgetting

On First-Order Meta-Learning Algorithms

Meta-Learning for Semi-Supervised Few-Shot Classification

A Generative Approach to Zero-Shot and Few-Shot Action Recognition

Moments in Time Dataset: One Million Videos for Event Understanding

Attribute-Enhanced Face Recognition with Neural Tensor Fusion Networks

Learning to Compare: Relation Network for Few-Shot Learning

Few-Shot Learning with Graph Neural Networks

Joint Dimensionality Reduction and Metric Learning: A Geometric Take

Meta-SGD: Learning to Learn Quickly for Few Shot Learning

Deep Co-occurrence Feature Learning for Visual Object Recognition

A simple neural network module for relational reasoning

Prototypical Networks for Few-shot Learning

Model-Agnostic Meta-Learning for Fast Adaptation of Deep Networks

Higher-Order Occurrence Pooling for Bags-of-Words: Visual Concept Detection

Domain Adaptation by Mixture of Alignments of Second-or Higher-Order Scatter Tensors

Generalized Deep Transfer Networks for Knowledge Propagation in Heterogeneous Domains

Semi-Supervised Classification with Graph Convolutional Networks

Matching Networks for One Shot Learning

Adversarial Feature Learning

Tensor Representations via Kernel Linearization for Action Recognition from 3D Skeletons

Learning Deep Features for Discriminative Localization

Weakly-Shared Deep Transfer Networks for Heterogeneous-Domain Knowledge Propagation

Bilinear CNN Models for Fine-Grained Visual Recognition

Going deeper with convolutions

Food-101 - Mining Discriminative Components with Random Forests

ImageNet Large Scale Visual Recognition Challenge

3D Object Representations for Fine-Grained Categorization