Efficient Augmentation

Efficient augmentation techniques aim to improve the performance of machine learning models by strategically modifying training data, addressing issues like data scarcity, class imbalance, and shortcut learning. Current research focuses on developing augmentation methods tailored to specific data types (e.g., time series, images, point clouds, event streams) and model architectures (e.g., contrastive learning, CLIP, Spiking Neural Networks), often incorporating principles like relevance propagation or contextual information to guide the augmentation process. These advancements enhance model generalization, reduce overfitting, and improve accuracy across diverse applications, ranging from grammatical error correction to semantic segmentation and object recognition.

Papers

October 31, 2024

A Novel Breast Ultrasound Image Augmentation Method Using Advanced Neural Style Transfer: An Efficient and Explainable Approach
Lipismita Panigrahi, Prianka Rani Saha, Jurdana Masuma Iqrah, Sushil Prasad
Deep Learning Data Augmentation Neural Style Transfer Breast Ultrasound Efficient Augmentation

July 12, 2024

Guidelines for Augmentation Selection in Contrastive Learning for Time Series Classification
Ziyu Liu, Azadeh Alavi, Minyi Li, Xiang Zhang
Contrastive Learning Time Series Classification State Aware Guideline Effective Augmentation Efficient Augmentation

June 25, 2024

Improving Grammatical Error Correction via Contextual Data Augmentation
Yixuan Wang, Baoxin Wang, Yijun Liu, Qingfu Zhu, Dayong Wu, Wanxiang Che
Data Augmentation Synthetic Data Generation Grammatical Error Correction Context Augmentation Efficient Augmentation

May 28, 2024

Learning to Detour: Shortcut Mitigating Augmentation for Weakly Supervised Semantic Segmentation
JuneHyoung Kwon, Eunju Lee, Yunsung Cho, YoungBin Kim
LeArning Abstract Semantic Segmentation Pixel Level Weakly Supervised Semantic Segmentation Background Generation Efficient Augmentation

March 14, 2024

EventRPG: Event Data Augmentation with Relevance Propagation Guidance
Mingyuan Sun, Donghao Zhang, Zongyuan Ge, Jiaxu Wang, Jia Li, Zheng Fang, Renjing Xu
Data Augmentation Spiking Neural Network Saliency Map Layer Wise Relevance Propagation Efficient Augmentation

November 28, 2023

Does VLN Pretraining Work with Nonsensical or Irrelevant Instructions?
Wang Zhu, Ishika Singh, Yuan Huang, Robin Jia, Jesse Thomason
Data Augmentation Human Instruction Vision and Language Navigation Language Navigation Pixel Trajectory Efficient Augmentation

April 18, 2023

METAM: Goal-Oriented Data Discovery
Sainyam Galhotra, Yue Gong, Raul Castro Fernandez
Causal Inference Task Efficient Augmentation Goal Discovery

December 16, 2022

Attentive Mask CLIP
Yifan Yang, Weiquan Huang, Yixuan Wei, Houwen Peng, Xinyang Jiang, Huiqiang Jiang, Fangyun Wei, Yin Wang, Han Hu, Lili Qiu, Yuqing Yang
Attention Mask Image Token Attention Masking Efficient Augmentation Dynamic Token

November 2, 2022

Joint Data and Feature Augmentation for Self-Supervised Representation Learning on Point Clouds
Zhuheng Lu, Yuewei Dai, Weiqing Li, Zhiyong Su
Point Cloud Data Augmentation Self Supervised Representation Learning Feature Augmentation Joint Modeling Efficient Augmentation Contrastive Augmentation

July 13, 2022

Efficient Augmentation for Imbalanced Deep Learning
Damien Dablain, Colin Bellinger, Bartosz Krawczyk, Nitesh Chawla
Convolutional Neural Network Imbalanced Data Generalization Gap Balanced Dataset Efficient Augmentation Deep Imbalanced

May 24, 2022

EventMix: An Efficient Augmentation Strategy for Event-Based Data
Guobin Shen, Dongcheng Zhao, Yi Zeng
Neuromorphic Computing Event Data Neuromorphic Datasets Event Stream Efficient Augmentation