Learning Data Manipulation for Augmentation and Weighting

By Zhiting Hu, Bowen Tan, Ruslan Salakhutdinov, Tom Michael Mitchell, Eric P. Xing. 2019

126 citations recorded by the source corpus.

Manipulating data, such as weighting data examples or augmenting with new instances, has been increasingly used to improve model training. Previous work has studied various rule- or learning-based approaches designed for specific types of data manipulation. In this work, we propose a new method that supports learning different manipulation schemes with the same gradient-based algorithm. Our approach builds upon a recent connection of supervised learning and reinforcement learning (RL), and adapts an off-the-shelf reward learning algorithm from RL for joint data manipulation learning and model training. Different parameterization of the ``data reward'' function instantiates different manipulation schemes. We showcase data augmentation that learns a text transformation network, and data weighting that dynamically adapts the data sample importance. Experiments show the resulting algorithms significantly improve the image and text classification performance in low data regime and class-imbalance problems.

Subfields: Artificial Intelligence, Optimization & Training

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

In Optimization & Training

Citations in the corpus

A Survey of Data Augmentation Approaches for NLP (2021, neutral): “, 2020b,c), and automated text augmentation (Hu et al., 2019; Cai et al., 2020).”
Good-Enough Compositional Data Augmentation (2019, background): “, 2017; Wei and Zhou, 2019); these approaches give improvements even when combined with large-scale pretraining (Hu et al., 2019).”
Data Augmentation using Pre-trained Transformer Models (2020, methodology): “These DA methods can be easily combined with other advances in text content manipulation such as co-training the data generator and classiﬁer (Hu et al., 2019).”
A Survey on Data Augmentation for Text Classification (2021, unknown): “Based on a CharCNN-LSTM on the AG News dataset, they achieve the best improvement in accuracy by obtaining an additional 0.62%.”
Continual Adaptation of Visual Representations via Domain Randomization and Meta-learning (2020, background): “For example, previous works focused on augmentation strategies to improve indomain [12, 25] and out-of-domain [61, 60, 47] performance might prove helpful in our settings.”
Pick and Choose: A GNN-based Imbalanced Learning Approach for Fraud Detection (2021, background): “Re-weighting methods assign different weights to different classes or even different samples by cost-sensitive adjustments [4, 9, 19, 21] or meta-learning based methods [14, 35, 37].”
AutoQA: From Databases to Q&A Semantic Parsers with Only Synthetic Training Data (2020, methodology): “We fine-tune it for 4 epochs on sentence pairs from PARABANK 2 (Hu et al., 2019a), which is a paraphrase dataset constructed by back-translating the Czech portion of an English-Czech parallel corpus.”
Data Augmentation Approaches in Natural Language Processing: A Survey (2021, background): “In addition to DA, there are some other data manipulation methods to improve model generalization [116, 92].”
Time Series Data Augmentation for Deep Learning: A Survey (2020, background): “A recent study investigates this topic in the area of CV and NLP [Hu et al. , 2019], which signiﬁcantly improves text and image classiﬁcation in low data regime and imbalanced class problems.”
An Empirical Survey of Data Augmentation for Limited Data Learning in NLP (2021, background): “Automatically learning data augmentation strategies or searching for an optimal augmentation policy for given datasets/tasks/models could enhance the generalizability of data augmentation techniques (Maharana and Bansal, 2020; Hu et al., 2019).”
CoDA: Contrast-enhanced and Diversity-promoting Data Augmentation for Natural Language Understanding (2020, background): “The discrete nature of natural language makes it challenging to design effective label-preserving transformations for text sequences that can help improve model generalization (Hu et al., 2019; Xie et al., 2019).”
On the Generalization Effects of Linear Transformations in Data Augmentation (2020, methodology): “%) use reinforcement learning based search methods (Hu et al., 2019).”
Bowtie Networks: Generative Modeling for Joint Few-Shot Recognition and Novel-View Synthesis (2020, background): “Joint Data Augmentation and Task Model Learning leverage generative networks to improve other visual tasks (Peng et al., 2018; Hu et al., 2019; Luo et al., 2020; Zhang et al., 2020).”