Data Loss

Data loss, a pervasive challenge across diverse scientific domains, significantly impacts the accuracy and reliability of analyses and predictions. Current research focuses on mitigating data loss through various strategies, including robust decentralized learning algorithms that maintain accuracy even with node or data disruptions, advanced imputation techniques like tensor decomposition for filling missing values in Non-Intrusive Load Monitoring, and novel denoising diffusion models for recovering lost power system measurements. These efforts are crucial for improving the trustworthiness of data-driven models in applications ranging from AI and energy management to predictive maintenance and epidemiological modeling.

Papers

May 3, 2024

Robustness of Decentralised Learning to Nodes and Data Disruption
Luigi Palmieri, Chiara Boldrini, Lorenzo Valerio, Andrea Passarella, Marco Conti, János Kertész
Native Robustness Decentralized Learning Artificial Intelligence Research Random Node Collective Learning Data Loss

April 28, 2024

A Note on Asynchronous Challenges: Unveiling Formulaic Bias and Data Loss in the Hayashi-Yoshida Estimator
Evangelos Georgiadis
Short Note Algorithmic Bias Linear Estimator Intrinsic Bias Asynchronous Learning Publication Bias Poisson Process Novel Estimator Data Loss

March 9, 2024

A PID-Controlled Non-Negative Tensor Factorization Model for Analyzing Missing Data in NILM
DengYu Shi
Matrix Factorization Tensor Decomposition Tensor Completion Non Intrusive Load Monitoring Missing Data Imputation Data Loss

December 7, 2023

Improved Efficient Two-Stage Denoising Diffusion Power System Measurement Recovery Against False Data Injection Attacks and Data Losses
Jianhua Pei, Jingyu Wang, Dongyuan Shi, Ping Wang
Diffusion Model Power System False Data Injection Attack Power System Measurement Signal Imputation Data Loss

September 6, 2023

TFBEST: Dual-Aspect Transformer with Learnable Positional Encoding for Failure Prediction
Rohan Mohapatra, Saptarshi Sengupta
Transformer Based Self Attention Positional Encoding Failure Prediction Remaining Useful Data Loss Disk Failure

February 3, 2023

PINN Training using Biobjective Optimization: The Trade-off between Data Loss and Residual Loss
Fabian Heldmann, Sarah Berkhahn, Matthias Ehrhardt, Kathrin Klamroth
Residual Stream Dynamical Regime Susceptible Infected Data Loss PINN Training

Data Loss

Papers

Robustness of Decentralised Learning to Nodes and Data Disruption

A Note on Asynchronous Challenges: Unveiling Formulaic Bias and Data Loss in the Hayashi-Yoshida Estimator

A PID-Controlled Non-Negative Tensor Factorization Model for Analyzing Missing Data in NILM

Improved Efficient Two-Stage Denoising Diffusion Power System Measurement Recovery Against False Data Injection Attacks and Data Losses

TFBEST: Dual-Aspect Transformer with Learnable Positional Encoding for Failure Prediction

PINN Training using Biobjective Optimization: The Trade-off between Data Loss and Residual Loss