Zero-Shot Learning
태스크의 하위 분류와 논문·코드·평가·기법·데이터셋을 탐색하세요.
- 하위 태스크
- 7
- 평가표
- 33
- 보존된 평가 기록
- 84
Zero-shot learning (ZSL) is a model's ability to detect classes never seen during training. The condition is that the classes are not known during supervised learning.
Earlier work in zero-shot learning use attributes in a two-step approach to infer unknown classes. In the computer vision context, more recent advances learn mappings from image feature space to semantic space. Other approaches learn non-linear multimodal embeddings. In the modern NLP context, language models can be evaluated on downstream tasks without fine tuning.
Benchmark datasets for zero-shot learning include aPY, AwA, and CUB, among others.
( Image credit: Prototypical Networks for Few shot Learning in PyTorch )
Further readings:
논문
이 태스크와 하위 분야의 관련 논문입니다. 실제 성능 비교는 평가 결과에서 확인하세요.
A robust and efficient video representation for action recognition
Learning discriminative trajectorylet detector sets for accurate skeleton-based action recognition
Temporal Localization of Fine-Grained Actions in Videos by Domain Transfer from Web Images
Label-Embedding for Image Classification
Discriminative Bayesian Dictionary Learning for Classification
Transductive Multi-label Zero-shot Learning
Transductive Multi-class and Multi-label Zero-shot Learning
Video-Based Action Recognition Using Rate-Invariant Analysis of Covariance Trajectories
Temporal Pyramid Pooling Based Convolutional Neural Networks for Action Recognition
Describing Videos by Exploiting Temporal Structure
Probabilistic Zero-shot Classification with Semantic Rankings
A new network-based algorithm for human activity recognition in video
Unsupervised Learning of Video Representations using LSTMs
Multi-Action Recognition via Stochastic Modelling of Optical Flow and Gradients
Semantic Embedding Space for Zero-Shot Action Recognition
Multimodal Task-Driven Dictionary Learning for Image Classification
Linear-time Online Action Detection From 3D Skeletal Data Using Bags of Gesturelets
Learning Latent Spatio-Temporal Compositional Model for Human Action Recognition
Feature Sampling Strategies for Action Recognition
Advances in Human Action Recognition: A Survey
Deep Convolutional Neural Networks for Action Recognition Using Depth Map Sequences
Transductive Multi-view Zero-Shot Learning
Reconstruction-free action inference from compressive imagers
From Visual Attributes to Adjectives through Decompositional Distributional Semantics
A Unified Perspective on Multi-Domain and Multi-Task Learning
Improving zero-shot learning by mitigating the hubness problem
Pooled Motion Features for First-Person Videos
Gradient Boundary Histograms for Action Recognition
Beyond Gaussian Pyramid: Multi-skip Feature Stacking for Action Recognition
Finding Action Tubes
A Nonparametric Bayesian Approach Toward Stacked Convolutional Independent Component Analysis
A Latent Clothing Attribute Approach for Human Pose Estimation
Zero-Shot Object Recognition System based on Topic Model
Evaluation of Output Embeddings for Fine-Grained Image Classification
Histogram of Oriented Principal Components for Cross-View Action Recognition
출처: Papers with Code · CC BY-SA 4.0. 아카이브 자료와 새로 수집한 논문은 별도로 관리합니다.