New Gradient-Enhanced Method Improves Online Estimation for High-Dimensional Machine Learning Models
Researchers have developed an improved renewable Lasso method for online estimation in high-dimensional generalized linear models that removes previous batch-number constraints. The approach uses a gradient-enhanced surrogate loss and extends to distributed streaming data across multiple sites. This advancement could improve the efficiency and scalability of machine learning systems processing continuous data streams.
A new machine learning method addresses limitations in existing renewable estimation approaches for high-dimensional generalized linear models with streaming data. The researchers propose a gradient-enhanced surrogate loss that approximates cumulative loss using only historical summaries, eliminating the batch-number constraints that restricted previous methods. The approach is extended to distributed settings where data batches are partitioned across multiple sites and only gradient vectors are exchanged between them, reducing computational burden on individual clients. Non-asymptotic error bounds are derived under high-dimensional scaling without stringent batch constraints. Simulation results on linear and logistic models, along with real-data applications, demonstrate improved accuracy compared to existing renewable estimators.
What different sources said
- arXiv cs.LGCenter
Renewable Lasso without Batch-Number Constraints: A Gradient-Enhanced Approach
Related
Genetic Drift, Not Selection, Drives Rapid Feather Color Evolution in Island Bird Radiation
A new study of an island bird radiation found that rapid evolution of feather coloration is driven primarily by genetic drift in small populations rather than sexual or ecological selection. The research integrated whole-genome data with detailed plumage measurements across complete species sampling to test whether signaling trait evolution correlates with speciation rates. The findings suggest that neutral demographic processes play a central role in generating phenotypic diversity during island radiations, challenging assumptions about the mechanisms driving rapid evolution.
New AI Model Improves Prediction of Therapeutic Peptide Function from Protein Sequences
Researchers developed a lightweight CNN classifier that predicts whether peptide sequences have therapeutic properties, trained on a database of 54,655 peptides across 48 functional categories. The model uses a novel negative sampling strategy to reduce false positive rates from over 60% in previous approaches to 2.1%. This advancement could accelerate drug discovery by enabling faster computational screening of peptide candidates before expensive experimental testing.
Study Shows Different Metabolic Stress Models Produce Distinct Effects on Human Neuronal Networks
Researchers tested three common in vitro metabolic stress models on human-derived neuronal networks and found each produced different patterns of neuronal activity and cell damage. The models tested were hypoxia alone, oxygen-glucose deprivation (OGD), and hypoxia combined with glutamate exposure. The findings suggest that choice of experimental model significantly affects results and that combining electrophysiological and structural analyses is important for accurately assessing metabolic stress in stroke research.