Training Data

Training data is crucial for machine learning model development, with current research focusing on improving data quality, efficiency, and mitigating biases. Active areas include generating synthetic data to address scarcity or privacy concerns, developing algorithms to optimize data selection and usage (e.g., self-paced learning, active learning), and mitigating issues like data contamination and imbalance through techniques such as data augmentation, selective parameter merging, and novel loss functions. The quality and characteristics of training data significantly impact model performance, generalization, and robustness, influencing various applications from natural language processing and image recognition to scientific computing and medical diagnosis.

Papers

October 18, 2023

Reflection-Tuning: Data Recycling Improves LLM Instruction-Tuning
Ming Li, Lichang Chen, Jiuhai Chen, Shwai He, Heng Huang, Jiuxiang Gu, Tianyi Zhou
Training Data Medical LLM Language Understanding Instruction Tuning LLM Output Data Recycling Physic Informed Tuning

October 17, 2023

QADYNAMICS: Training Dynamics-Driven Synthetic QA Diagnostic for Zero-Shot Commonsense Question Answering
Haochen Shi, Weiqi Wang, Tianqing Fang, Baixuan Xu, Wenxuan Ding, Xin Liu, Yangqiu Song
Language Model Training Data Commonsense Knowledge Commonsense Question Answering Synthetic Question Zero Shot Commonsense

October 16, 2023

October 14, 2023

Plug-and-Play Feature Generation for Few-Shot Medical Image Classification
Qianyu Guo, Huifang Du, Xing Jia, Shuyong Gao, Yan Teng, Haofen Wang, Wenqiang Zhang
Training Data Medical Image Shot Learning Medical Image Classification Feature Generation Effective Feature

October 13, 2023

October 9, 2023

October 8, 2023

On the Zero-Shot Generalization of Machine-Generated Text Detectors
Xiao Pu, Jingyu Zhang, Xiaochuang Han, Yulia Tsvetkov, Tianxing He
Large Language Model Zero Shot Training Data Machine Generated Generated Text Machine Generated Text

October 6, 2023

From Zero to Hero: Detecting Leaked Data through Synthetic Data Injection and Model Querying
Biao Wu, Qiang Huang, Anthony K. H. Tung
Training Data Machine Learning Application Better Zero Real World Datasets Zero to HeRo Model Join

October 5, 2023

Improving classifier decision boundaries using nearest neighbors
Johannes Schneider
Adversarial Attack Training Data Nearest Neighbor Decision Boundary Pre Trained Convolutional Neural Network

October 4, 2023

October 3, 2023

Secure and Effective Data Appraisal for Machine Learning
Xu Ouyang, Changhong Yang, Felix Xiaozhu Lin, Yangfeng Ji
Machine Learning Training Data Data Selection Data Market MLP Architecture Party Computation

Training Data

Papers

Reflection-Tuning: Data Recycling Improves LLM Instruction-Tuning

QADYNAMICS: Training Dynamics-Driven Synthetic QA Diagnostic for Zero-Shot Commonsense Question Answering

Data Contamination Through the Lens of Time

Repetition In Repetition Out: Towards Understanding Neural Text Degeneration from the Data Perspective

Plug-and-Play Feature Generation for Few-Shot Medical Image Classification

PuoBERTa: Training and evaluation of a curated language model for Setswana

Training and Predicting Visual Error for Real-Time Applications

Federated Meta-Learning for Few-Shot Fault Diagnosis with Representation Encoding

Dynamic Sparse No Training: Training-Free Fine-tuning for Sparse LLMs

Strategies and impact of learning curve estimation for CNN-based image classification

Debias the Training of Diffusion Models

Effects of Human Adversarial and Affable Samples on BERT Generalization

An operator preconditioning perspective on training in physics-informed machine learning

Unlearning with Fisher Masking

On the Zero-Shot Generalization of Machine-Generated Text Detectors

From Zero to Hero: Detecting Leaked Data through Synthetic Data Injection and Model Querying

Improving classifier decision boundaries using nearest neighbors

Attributing Learned Concepts in Neural Networks to Training Data

Out-of-Distribution Detection by Leveraging Between-Layer Transformation Smoothness

Secure and Effective Data Appraisal for Machine Learning