45,292 Hits in 4.4 sec

Automatic Data Augmentation for Generalization in Deep Reinforcement Learning [article]

Roberta Raileanu, Max Goldstein, Denis Yarats, Ilya Kostrikov, Rob Fergus
2021 arXiv   pre-print
In this paper, we compare three approaches for automatically finding an appropriate augmentation.  ...  Deep reinforcement learning (RL) agents often fail to generalize to unseen scenarios, even when they are trained on many instances of semantically similar environments.  ...  Automatic Data Augmentation for RL Data Augmentation in RL Image augmentation has been successfully applied in computer vision for improving generalization on object classification tasks (Simard et  ... 
arXiv:2006.12862v2 fatcat:cbrw6bas4rajdf2usv3tqlslha

Automatic Data Augmentation via Deep Reinforcement Learning for Effective Kidney Tumor Segmentation [article]

Tiexin Qin and Ziyuan Wang and Kelei He and Yinghuan Shi and Yang Gao and Dinggang Shen
2020 arXiv   pre-print
In this paper, we developed a novel automatic learning-based data augmentation method for medical image segmentation which models the augmentation task as a trial-and-error procedure using deep reinforcement  ...  Conventional data augmentation realized by performing simple pre-processing operations (\eg, rotation, crop, \etc) has been validated for its advantage in enhancing the performance for medical image segmentation  ...  Regarding this challenge, we in this paper present a learning-based automatic data augmentation method for medical image segmentation, which introduces the deep reinforcement learning (DRL) to explore  ... 
arXiv:2002.09703v1 fatcat:ilt6yxnybndwnlqrs5t4sygete

Automatic Data Augmentation by Learning the Deterministic Policy [article]

Yinghuan Shi, Tiexin Qin, Yong Liu, Jiwen Lu, Yang Gao, Dinggang Shen
2019 arXiv   pre-print
Aiming to produce sufficient and diverse training samples, data augmentation has been demonstrated for its effectiveness in training deep models.  ...  Specifically, the current augmented subset is required to maximize the performance improvement compared with the last augmented subset by learning the deterministic augmentation policy using deep reinforcement  ...  Also, we will exploit the different action selection preferences in different datasets to guide better action design and initialization.  ... 
arXiv:1910.08343v2 fatcat:kbmxddhnqbca3efmmcbs7rrnx4

Learning to Augment for Data-Scarce Domain BERT Knowledge Distillation [article]

Lingyun Feng, Minghui Qiu, Yaliang Li, Hai-Tao Zheng, Ying Shen
2021 arXiv   pre-print
To tackle this problem, we propose a method to learn to augment for data-scarce domain BERT knowledge distillation, by learning a cross-domain manipulation scheme that automatically augments the target  ...  Specifically, the proposed method generates samples acquired from a stationary distribution near the target data and adopts a reinforced selector to automatically refine the augmentation strategy according  ...  Unlike manually designed data augmentation methods for in-domain data, our model can learn cross-domain manipulations and automatically augment data according to the feedback of the student model.  ... 
arXiv:2101.08106v2 fatcat:r522vriyc5eeramlnb5esa23lm

Automatic Data Augmentation for 3D Medical Image Segmentation [article]

Ju Xu, Mengzhang Li, Zhanxing Zhu
2020 arXiv   pre-print
Data augmentation is an effective and universal technique for improving generalization performance of deep neural networks.  ...  To fully exploit the potential of data augmentation, we propose an efficient algorithm to automatically search for the optimal augmentation strategies.  ...  In [16] , the authors proposed to utilize reinforcement learning to search for augmentation strategies.  ... 
arXiv:2010.11695v2 fatcat:6d4ov4hhdngijbbpyvtmjm3eyi

Synthetic Sample Selection via Reinforcement Learning [article]

Jiarong Ye, Yuan Xue, L. Rodney Long, Sameer Antani, Zhiyun Xue, Keith Cheng, Xiaolei Huang
2020 arXiv   pre-print
However, the quality control of synthetic images for data augmentation purposes is under-investigated, and some of the generated images are not realistic and may contain misleading features that distort  ...  In this work, we propose a reinforcement learning (RL) based synthetic sample selection method that learns to choose synthetic images containing reliable and informative features.  ...  Joe Stanley of Missouri University of Science and Technology that made the histopathology data collection possible.  ... 
arXiv:2008.11331v1 fatcat:g6f6vfqflrcklmkzflihakx4pi

Using Semantic Information to Improve Generalization of Reinforcement Learning Policies for Autonomous Driving

Florence Carton, David Filliat, Jaonary Rabarisoa, Quoc Cuong Pham
2021 2021 IEEE Winter Conference on Applications of Computer Vision Workshops (WACVW)  
To improve generalization, data augmentation is a widely used practice in deep learning. However even if it is unavoidable in supervised learning, it is less the case in reinforcement learning.  ...  This data augmentation is inspired from the one used in the original CARLA paper for Figure 4 : Examples of data augmentation the supervised learning training [9] .  ...  Hyperparameters for Reinforcement Learning Training In reinforcement learning, hyperparameters tuning is crucial.  ... 
doi:10.1109/wacvw52041.2021.00020 fatcat:ovkmnoa2ljdddgep5l5nq53xk4

Label Augmentation with Reinforced Labeling for Weak Supervision [article]

Gürkan Solmaz, Flavio Cirillo, Fabio Maresca, Anagha Gode Anil Kumar
2022 arXiv   pre-print
This is due to the common data programming pipeline that neglects to utilize data features during the generative process. This paper proposes a new approach called reinforced labeling (RL).  ...  Weak supervision (WS) is an alternative to the traditional supervised learning to address the need for ground truth.  ...  Different methods can utilize data features in the generative process for augmenting the labels in the labeling matrix.  ... 
arXiv:2204.06436v1 fatcat:vrikjhjz35fvhffxy7cvativdm

Sampling Strategies for GAN Synthetic Data [article]

Binod Bhattarai, Seungryul Baek, Rumeysa Bodur, Tae-Kyun Kim
2019 arXiv   pre-print
In addition to this, we explore reinforcement learning (RL) to automatically search a subset of meaningful synthetic examples from a large pool of GAN synthetic data.  ...  This data is being utilized for augmenting with real examples in order to train deep Convolutional Neural Networks (CNNs).  ...  Please note that our methods are generic and can be applied for VAEs synthetic data too. Reinforcement Learning.  ... 
arXiv:1909.04689v1 fatcat:q5rqagm5ere5hm2stqmlaqienq

Machine Reading Comprehension Framework Based on Self-training for Domain Adaptation

Hyeon-gu Lee, Youngjin Jang, Harksoo Kim
2021 IEEE Access  
on automatic data augmentation.  ...  During the mutual self-training, the pseudo-question generator provides new training data to the MRC system and obtains rewards from the MRC system for reinforcement learning.  ...  Through reinforcement learning [1] based on the evaluation of the MRC model, the question generator improves its performances and augments the training data for the target domain.  ... 
doi:10.1109/access.2021.3054912 fatcat:5vjfedeqdncy5komjdm5oik6ta

Dialog State Tracking with Reinforced Data Augmentation [article]

Yichun Yin, Lifeng Shang, Xin Jiang, Xiao Chen, Qun Liu
2019 arXiv   pre-print
In this paper, we address this difficulty by proposing a reinforcement learning (RL) based framework for data augmentation that can generate high-quality data to improve the neural state tracker.  ...  Specifically, we introduce a novel contextual bandit generator to learn fine-grained augmentation policies that can generate new effective instances by choosing suitable replacements for the specific context  ...  In contrast, we propose a coarseto-fine strategy for data augmentation, where the fine-grained generative polices learned by RL are used to automatically reduce the noisy instances and retain the effective  ... 
arXiv:1908.07795v2 fatcat:hunfhyvehrcadmubo3wz5q4uua

Deep Reinforcement Learning with Mixed Convolutional Network [article]

Yanyu Zhang
2020 arXiv   pre-print
The dataset is generated by playing the game manually in Gym and used a data augmentation method to expand the dataset to 4 times larger than before.  ...  This paper presents a convolutional neural network (CNN) to playing the CarRacing-v0 using imitation learning in OpenAI Gym.  ...  Introduction OpenAI Gym is a popular open-source repository of reinforcement learning (RL) ( [5] [8] ) environments and development tools.  ... 
arXiv:2010.00717v2 fatcat:4aecue2qfzgstea5up763e73iu

Dialog State Tracking with Reinforced Data Augmentation

Yichun Yin, Lifeng Shang, Xin Jiang, Xiao Chen, Qun Liu
In this paper, we address this difficulty by proposing a reinforcement learning (RL) based framework for data augmentation that can generate high-quality data to improve the neural state tracker.  ...  Specifically, we introduce a novel contextual bandit generator to learn fine-grained augmentation policies that can generate new effective instances by choosing suitable replacements for specific context  ...  Acknowledgments We thank Milan Gritta for his fruitful discussions and all anonymous reviewer for their insightful comments.  ... 
doi:10.1609/aaai.v34i05.6491 fatcat:zfzv2mogjjewnoukemvst5rvii

Automated Augmentation with Reinforcement Learning and GANs for Robust Identification of Traffic Signs using Front Camera Images [article]

Sohini Roy Chowdhury, Lars Tornberg, Robin Halvfordsson, Jonatan Nordh, Adam Suhren Gustafsson, Joel Wall, Mattias Westerberg, Adam Wirehed, Louis Tilloy, Zhanying Hu, Haoyuan Tan, Meng Pan, Jonas Sjoberg
2019 arXiv   pre-print
In this work, we present an end-to-end framework to augment traffic sign training data using optimal reinforcement learning policies and a variety of Generative Adversarial Network (GAN) models, that can  ...  As a solution, machine learning models must be trained with data from multiple domains, and collecting and labeling more data in each new domain is time consuming and expensive.  ...  Reinforcement Learning based Augmentation The RL based data augmentation method (RLAUG) in [8] automatically searches for image processing policies or operations that can improve ODS performance by data  ... 
arXiv:1911.06486v1 fatcat:wcuk4zla35fy3eachacghb24fq

LexicalAT: Lexical-Based Adversarial Reinforcement Training for Robust Sentiment Classification

Jingjing Xu, Liang Zhao, Hanqi Yan, Qi Zeng, Yun Liang, Xu SUN
2019 Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing and the 9th International Joint Conference on Natural Language Processing (EMNLP-IJCNLP)  
Due to the discrete generation step in the generator, we use policy gradient, a reinforcement learning approach, to train the two modules.  ...  The proposed approach consists of a generator and a classifier. The generator learns to generate examples to attack the classifier while the classifier learns to defend these attacks.  ...  Acknowledgments We thank all reviewers for providing the thoughtful and constructive suggestions. This work was supported in part by National Natural Science Foundation of China (No. 61673028).  ... 
doi:10.18653/v1/d19-1554 dblp:conf/emnlp/XuZYZLS19 fatcat:czux2pcmp5dwdczl4t7xbt2c5e
« Previous Showing results 1 — 15 out of 45,292 results