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

June 5, 2024

Interactive Image Selection and Training for Brain Tumor Segmentation Network
Matheus A. Cerqueira, Flávia Sprenger, Bernardo C. A. Teixeira, Alexandre X. Falcão
Deep Learning Training Data Brain Tumor Segmentation Convolutional Layer Training Image U Shaped Network
A Human-Annotated Video Dataset for Training and Evaluation of 360-Degree Video Summarization Methods
Ioannis Kontostathis, Evlampios Apostolidis, Vasileios Mezaris
Training Data Global Evaluation Video Summarization Video Annotation Video Content State of the Art Summarization Ground Truth Summary

June 4, 2024

Astral: training physics-informed neural networks with error majorants
Vladimir Fanaskov, Tianchi Yu, Alexander Rudikov, Ivan Oseledets
Training Data Loss Function Physic Informed Neural Network Error Feedback Residual Stream Physic Informed Learning
Multiway Multislice PHATE: Visualizing Hidden Dynamics of RNNs through Training
Jiancheng Xie, Lou C. Kohler Voinov, Noga Mudrik, Gal Mishne, Adam Charles
Training Data Recurrent Neural Network \Sigma}{\Delta}$ Low Pas RNN Network Representation Hidden Dynamic Multi Slice

June 3, 2024

Model for Peanuts: Hijacking ML Models without Training Access is Possible
Mahmoud Ghorbel, Halima Bouzidi, Ioan Marius Bilasco, Ihsen Alouani
Training Data Machine Learning Model Full Model Hijacking Task Model Hijacking Attack
SAVA: Scalable Learning-Agnostic Data Valuation
Samuel Kessler, Tam Le, Vu Nguyen
Training Data Large Datasets Data Valuation
Cohort Squeeze: Beyond a Single Communication Round per Cohort in Cross-Device Federated Learning
Kai Yi, Timur Kharisov, Igor Sokolov, Peter Richtárik
Training Data Client Selection FedAvg Converges Cross Device Federated Learning Expert Delineated Cohort Communication Round
Frequency Enhanced Pre-training for Cross-city Few-shot Traffic Forecasting
Zhanyu Liu, Jianrong Ding, Guanjie Zheng
Training Data Traffic Flow Forecasting Frequency Enhancement Cross City Few Shot
Advancing DRL Agents in Commercial Fighting Games: Training, Integration, and Agent-Human Alignment
Chen Zhang, Qiang He, Zhou Yuan, Elvis S. Liu, Hong Wang, Jian Zhao, Yang Wang
Deep Reinforcement Learning Training Data Agent Alignment Fighting Game

May 31, 2024

How In-Context Learning Emerges from Training on Unstructured Data: On the Role of Co-Occurrence, Positional Information, and Noise Structures
Kevin Christian Wibisono, Yixin Wang
Large Language Model Training Data Context Learning Unstructured Data Co Occurrence Semi Structured Unstructured Text Data Positional Information Unsupervised Training
Training on the Edge of Stability Is Caused by Layerwise Jacobian Alignment
Mark Lowell, Catharine Kastner
Training Data Extreme Edge Core Stability Gradient Flow Hessian Matrix Gradient Descent Dynamic
Latent Intrinsics Emerge from Training to Relight
Xiao Zhang, William Gao, Seemandhar Jain, Michael Maire, David. A. Forsyth, Anand Bhattad
Training Data Human Relighting Inverse Graphic
Conditioning GAN Without Training Dataset
Kidist Amde Mekonnen
Training Data GAN Model Deep Learning Algorithm Generative Neural Network Conditional GANs GAN Algorithm

May 30, 2024

Mitigating the Impact of Labeling Errors on Training via Rockafellian Relaxation
Louis L. Chen, Bobbie Chern, Eric Eckstrand, Amogh Mahapatra, Johannes O. Royset
Training Data Global Impact Noisy Label Weak Label Labeling Error
SPOT: Text Source Prediction from Originality Score Thresholding
Edouard Yvinec, Gabriel Kasser
Large Language Model Training Data Text Modality SpOT Robot Originality Score Text Source
Improving the Training of Rectified Flows
Sangyun Lee, Zinan Lin, Giulia Fanti
Training Data Consistency Training Consistency Distillation Progressive Distillation Distillation Method Rectified Flow
Can the accuracy bias by facial hairstyle be reduced through balancing the training data?
Kagan Ozturk, Haiyu Wu, Kevin W. Bowyer
Training Data Face Recognition Model Facial Hair
Improving Object Detector Training on Synthetic Data by Starting With a Strong Baseline Methodology
Frank A. Ruis, Alma M. Liezenga, Friso G. Heslinga, Luca Ballan, Thijs A. Eker, Richard J. M. den Hollander, Martin C. van Leeuwen, Judith Dijk, Wyke Huizinga
Training Data Synthetic Data Object Detector Object Detection Model Strong Baseline Pre Trained Object

May 29, 2024

Critical Learning Periods: Leveraging Early Training Dynamics for Efficient Data Pruning
Everlyn Asiko Chimoto, Jay Gala, Orevaoghene Ahia, Julia Kreutzer, Bruce A. Bassett, Sara Hooker
Training Data Training Dynamic Neural Machine Translation Model Data Pruning Critical Period
Deep Positive-Unlabeled Anomaly Detection for Contaminated Unlabeled Data
Hiroshi Takahashi, Tomoharu Iwata, Atsutoshi Kumagai, Yuuki Yamanaka
Training Data Anomaly Score Positive Unlabeled Learning Semi Supervised Anomaly Detection Anomaly Detector Anomaly Data

Training Data

Papers

Interactive Image Selection and Training for Brain Tumor Segmentation Network

A Human-Annotated Video Dataset for Training and Evaluation of 360-Degree Video Summarization Methods

Astral: training physics-informed neural networks with error majorants

Multiway Multislice PHATE: Visualizing Hidden Dynamics of RNNs through Training

Model for Peanuts: Hijacking ML Models without Training Access is Possible

SAVA: Scalable Learning-Agnostic Data Valuation

Cohort Squeeze: Beyond a Single Communication Round per Cohort in Cross-Device Federated Learning

Frequency Enhanced Pre-training for Cross-city Few-shot Traffic Forecasting

Advancing DRL Agents in Commercial Fighting Games: Training, Integration, and Agent-Human Alignment

How In-Context Learning Emerges from Training on Unstructured Data: On the Role of Co-Occurrence, Positional Information, and Noise Structures

Training on the Edge of Stability Is Caused by Layerwise Jacobian Alignment

Latent Intrinsics Emerge from Training to Relight

Conditioning GAN Without Training Dataset

Mitigating the Impact of Labeling Errors on Training via Rockafellian Relaxation

SPOT: Text Source Prediction from Originality Score Thresholding

Improving the Training of Rectified Flows

Can the accuracy bias by facial hairstyle be reduced through balancing the training data?

Improving Object Detector Training on Synthetic Data by Starting With a Strong Baseline Methodology

Critical Learning Periods: Leveraging Early Training Dynamics for Efficient Data Pruning

Deep Positive-Unlabeled Anomaly Detection for Contaminated Unlabeled Data