Zero-Shot Recognition via Direct Classifier Learning with Transferred Samples and Pseudo Labels

As an interesting and emerging topic, zero-shot recognition (ZSR) makes it possible to train a recognition model by specifying the category's attributes when there are no labeled exemplars available. The fundamental idea for ZSR is to transfer knowledge from the abundant labeled data in different but related source classes via the class attributes. Conventional ZSR approaches adopt a two-step strategy in test stage, where the samples are projected into the attribute space in the first step, and then the recognition is carried out based on considering the relationship between samples and classes in the attribute space. Due to this intermediate transformation, information loss is unavoidable, thus degrading the performance of the overall system. Rather than following this two-step strategy, in this paper, we propose a novel one-step approach that is able to perform ZSR in the original feature space by using directly trained classifiers. To tackle the problem that no labeled samples of target classes are available, we propose to assign pseudo labels to samples based on the reliability and diversity, which in turn will be used to train the classifiers. Moreover, we adopt a robust SVM that accounts for the unreliability of pseudo labels. Extensive experiments on four datasets demonstrate consistent performance gains of our approach over the state-of-the-art two-step ZSR approaches.

• Publication year

  2017

• Venue

  AAAI Conference on Artificial Intelligence

• Publication date

  2017-02-12

• Fields of study

  Computer Science

• Identifiers
  DOI 10.1609/aaai.v31i1.11191
• External record

  Open on Semantic Scholar

• Source metadata

  Semantic Scholar

• No claims are published for this paper.

• No concepts are published for this paper.

• Transductive Zero-Shot Recognition via Shared Model Space Learning

  2016influential reference
• Recovering the Missing Link: Predicting Class-Attribute Associations for Unsupervised Zero-Shot Learning

  2016cited by this paper
• Synthesized Classifiers for Zero-Shot Learning

  2016cited by this paper
• Semi-Supervised Zero-Shot Classification with Label Representation Learning

  2015cited by this paper
• Unsupervised Domain Adaptation for Zero-Shot Learning

  2015cited by this paper
• An embarrassingly simple approach to zero-shot learning

  2015cited by this paper
• Max-Margin Zero-Shot Learning for Multi-class Classification

  2015cited by this paper
• Zero-shot object recognition by semantic manifold distance

  2015cited by this paper
• Zero-Shot Learning via Semantic Similarity Embedding

  2015influential reference
• Zero-shot recognition with unreliable attributes

  2014cited by this paper
• Transductive Multi-view Embedding for Zero-Shot Recognition and Annotation

  2014cited by this paper
• Alternatively Constrained Dictionary Learning For Image Superresolution

  2014cited by this paper
• Zero-Shot Learning Through Cross-Modal Transfer

  2013influential reference
• Label-Embedding for Attribute-Based Classification

  2013cited by this paper
• Zero-Shot Learning by Convex Combination of Semantic Embeddings

  2013cited by this paper
• Adversarial Label Flips Attack on Support Vector Machines

  2012cited by this paper
• ImageNet classification with deep convolutional neural networks

  2012cited by this paper
• Improving Word Representations via Global Context and Multiple Word Prototypes

  2012cited by this paper
• SUN attribute database: Discovering, annotating, and recognizing scene attributes

  2012cited by this paper
• LIBSVM: A library for support vector machines

  2011cited by this paper
• Describing objects by their attributes

  2009cited by this paper
• Learning to detect unseen object classes by between-class attribute transfer

  2009cited by this paper
• Learning Multiple Layers of Features from Tiny Images

  2009influential reference
• Nearly-linear time algorithms for graph partitioning, graph sparsification, and solving linear systems

  2003cited by this paper
• Laplacian Eigenmaps and Spectral Techniques for Embedding and Clustering

  2001cited by this paper
• Support-Vector Networks

  1995cited by this paper
• Attribute-Based Classification for Zero-Shot Visual Object Categorization

  year unknowninfluential reference

• Transductive zero-shot learning via knowledge graph and graph convolutional networks

  2025cites this paper
• Iterative Class Prototype Calibration for Transductive Zero-Shot Learning

  2023cites this paper
• Zero-shot learning and its applications from autonomous vehicles to COVID-19 diagnosis: A review

  2020influential citation
• Bidirectional generative transductive zero-shot learning

  2020cites this paper
• Attribute-Induced Bias Eliminating for Transductive Zero-Shot Learning

  2020cites this paper
• Transductive Zero-Shot Hashing for Multilabel Image Retrieval

  2019cites this paper
• Triple Verification Network for Generalized Zero-Shot Learning

  2019cites this paper
• Sensor-based human activity recognition via zero-shot learning

  2019cites this paper
• A Survey of Zero-Shot Learning

  2019influential citation
• Progressive Ensemble Networks for Zero-Shot Recognition

  2018influential citation
• Self-Training Ensemble Networks for Zero-Shot Image Recognition

  2018influential citation
• MSplit LBI: Realizing Feature Selection and Dense Estimation Simultaneously in Few-shot and Zero-shot Learning

  2018cites this paper
• Safe semi supervised multi-target regression (MTR-SAFER) for new targets learning

  2018cites this paper
• Zero-Shot Learning With Attribute Selection

  2018cites this paper
• Zero-shot image classification

  2017cites this paper
• Synthesizing Samples for Zero-shot Learning

  2017cites this paper
• Elephant Panda Lion Dolphin Dog Monkey ... ... Classes Elephant Dolphin Lion Classifier Training Data Synthesizing Prediction

  2017cites this paper
• Zero-shot Learning via Shared-Reconstruction-Graph Pursuit

  2017cites this paper