Core Set

Coresets are strategically selected subsets of a larger dataset that effectively represent the entire dataset's characteristics for various machine learning tasks. Current research focuses on improving coreset selection algorithms, often incorporating density awareness and diversity metrics to optimize for both computational efficiency and model accuracy in applications ranging from text processing and image classification to scientific simulations. This approach offers significant advantages by reducing computational costs and data annotation burdens, leading to faster and more efficient model training while maintaining or even improving performance. The impact spans diverse fields, enabling more scalable and cost-effective solutions for large-scale data analysis and machine learning applications.

Papers

March 18, 2024

The Power of Few: Accelerating and Enhancing Data Reweighting with Coreset Selection
Mohammad Jafari, Yimeng Zhang, Yihua Zhang, Sijia Liu
Real Power Robust Representation Large Datasets Model Accuracy Machine Learning Task Data Reweighting Core Set

December 15, 2023

Density Matters: Improved Core-set for Active Domain Adaptive Segmentation
Shizhan Liu, Zhengkai Jiang, Yuxi Li, Jinlong Peng, Yabiao Wang, Weiyao Lin
Semantic Segmentation Greedy Algorithm Particle Density Active Domain Adaptation Core Set

October 25, 2023

DEFT: Data Efficient Fine-Tuning for Pre-Trained Language Models via Unsupervised Core-Set Selection
Devleena Das, Vivek Khetan
Pre Trained Language Model Tuned Lm Editing Model Core Set Data Efficient Fine Tuning

October 12, 2023

Core-sets for Fair and Diverse Data Summarization
Sepideh Mahabadi, Stojan Trajanovski
Core Set Diversity Maximization Table Summarization

September 15, 2023

Diversity-based core-set selection for text-to-speech with linguistic and acoustic features
Kentaro Seki, Shinnosuke Takamichi, Takaaki Saeki, Hiroshi Saruwatari
Text to Speech Acoustic Feature Phoneme Recognition Core Set

August 10, 2023

Composable Core-sets for Diversity Approximation on Multi-Dataset Streams
Stephanie Wang, Michael Flynn, Fangyu Luo
Gradient Descent Core Set

June 10, 2022

In Defense of Core-set: A Density-aware Core-set Selection for Active Learning
Yeachan Kim, Bonggun Shin
Active Learning Particle Density Core Set

May 28, 2022

Core-set Selection Using Metrics-based Explanations (CSUME) for multiclass ECG
Sagnik Dakshit, Barbara Mukami Maweu, Sristi Dakshit, Balakrishnan Prabhakaran
Deep Learning Evaluation Metric Decision Support System Electrocardiogram Data Core Set

March 27, 2022

Pyramid-BERT: Reducing Complexity via Successive Core-set based Token Selection
Xin Huang, Ashish Khetan, Rene Bidart, Zohar Karnin
NLP Task Complexity Matter Transformer Based Language Model Code BERT Core Set Token Selection BERT LID Model

February 9, 2022

Improving greedy core-set configurations for active learning with uncertainty-scaled distances
Yuchen Li, Frank Rudzicz
Active Learning Image Classification Model Uncertainty Core Set Quadratic Speedup

November 15, 2021

Reducing the Long Tail Losses in Scientific Emulations with Active Learning
Yi Heng Lim, Muhammad Firmansyah Kasim
Deep Learning Model Active Learning Model Performance Core Set Tailed Loss Scientific Simulation