Temporal Ensembling for Semi-Supervised Learning
Explore this paper's citation graph
Summary
Self-ensembling is introduced, where it is shown that this ensemble prediction can be expected to be a better predictor for the unknown labels than the output of the network at the most recent training epoch, and can thus be used as a target for training.
- Type
- preprint
- Published
- 2016-10-07
- Cited by
- 2,932
- References
- 42
- Access
- Open access
- OpenAlex
- https://openalex.org/W2951970475
- Semantic Scholar
- https://api.semanticscholar.org/CorpusID:13123084
Keywords
Computer science, Regularization (linguistics), Artificial intelligence, Machine learning, Training set
References
- Regularization of Neural Networks using DropConnect
- Lasagne: First release.
- Learning from labeled and unlabeled data with label propagation
- Delving Deep into Rectifiers: Surpassing Human-Level Performance on ImageNet Classification
- Distilling the Knowledge in a Neural Network
- The conference paper
- Dropout: a simple way to prevent neural networks from overfitting
- Bootstrapping via Graph Propagation
- Unsupervised Word Sense Disambiguation Rivaling Supervised Methods
- Semi-supervised Learning with Deep Generative Models
- Learning with Pseudo-Ensembles
- Fractional Max-Pooling
- 80 Million Tiny Images: A Large Data Set for Nonparametric Object and Scene Recognition
- Theano: A Python framework for fast computation of mathematical expressions
- Mutual exclusivity loss for semi-supervised deep learning
- Semi-Supervised Learning with Generative Adversarial Networks
- Improved Techniques for Training GANs
- Making Deep Neural Networks Robust to Label Noise: A Loss Correction Approach
- Semi-supervised Learning with Ladder Networks
- Training Deep Neural Networks on Noisy Labels with Bootstrapping
Cited by
- Semisupervised Text Classification by Variational Autoencoder
- Triple Generative Adversarial Nets
- Virtual Adversarial Training: A Regularization Method for Supervised and Semi-Supervised Learning
- GAR: An efficient and scalable Graph-based Activity Regularization for semi-supervised learning
- Self-ensembling for domain adaptation
- Weight-averaged consistency targets improve semi-supervised deep learning results
- Adversarial Dropout for Supervised and Semi-supervised Learning
- Rocket Launching: A Universal and Efficient Framework for Training Well-performing Light Net
- PixelGAN Autoencoders
- VisDA: The Visual Domain Adaptation Challenge
- Self-ensembling for visual domain adaptation
- Multi-modal emotion recognition using semi-supervised learning and multiple neural networks in the wild
- Virtual Adversarial Ladder Networks For Semi-supervised Learning
- Multiplicative Noise Channel in Generative Adversarial Networks
- Hybrid Deep Discriminative/Generative Models for Semi-Supervised Learning.
- Exploiting Target Data to Learn Deep Convolutional Networks for Scene-Adapted Human Detection
- Semi-Supervised and Active Few-Shot Learning with Prototypical Networks
- Semi-Supervised Learning with IPM-based GANs: an Empirical Study
- A Semi-Supervised Two-Stage Approach to Learning from Noisy Labels
- A DIRT-T Approach to Unsupervised Domain Adaptation
Related papers
- On two recent nonconvex penalties for regularization in machine learning
- Self-Training with Selection-by-Rejection
- Universum Prescription: Regularization Using Unlabeled Data
- Automatic Labeling of Tweets for Crisis Response Using Distant Supervision
- Enhancing Self-Training Methods
- A new cross-training approach by using labeled data
- A Tri-training Based Transfer Learning Algorithm
- Semi-Supervised Learning for Natural Language Processing