Learning to Learn from Weak Supervision by Full Supervision

Abstract

In this paper, we propose a method for training neural networks when we have a large set of data with weak labels and a small amount of data with true labels. In our proposed model, we train two neural networks: a target network, the learner and a confidence network, the meta-learner. The target network is optimized to perform a given task and is trained using a large set of unlabeled data that are weakly annotated. We propose to control the magnitude of the gradient updates to the target network using the scores provided by the second confidence network, which is trained on a small amount of supervised data. Thus we avoid that the weight updates computed from noisy labels harm the quality of the target network model.

Accepted at NIPS Workshop on Meta-Learning (MetaLearn 2017), Long Beach, CA, USA

References in corpus (3)

• Adam: A Method for Stochastic Optimization
• Distilling the Knowledge in a Neural Network
• TF.Learn: TensorFlow's High-level Module for Distributed Machine Learning

Cited by in corpus (12)

• Meta-Weight-Net: Learning an Explicit Mapping For Sample Weighting
• Image Classification with Deep Learning in the Presence of Noisy Labels: A Survey
• Avoiding Your Teacher's Mistakes: Training Neural Networks with Controlled Weak Supervision
• Fast and Three-rious: Speeding Up Weak Supervision with Triplet Methods
• Learning Fast Sample Re-weighting Without Reward Data
• Fidelity-Weighted Learning
• Learning More From Less: Towards Strengthening Weak Supervision for Ad-Hoc Retrieval
• Multi-Resolution Weak Supervision for Sequential Data
• Train and You'll Miss It: Interactive Model Iteration with Weak Supervision and Pre-Trained Embeddings
• Semi-Supervised Confidence Network aided Gated Attention based Recurrent Neural Network for Clickbait Detection
• Robust Collaborative Learning with Noisy Labels
• Learning to Rank from Samples of Variable Quality