A Bayesian Framework for Built-in Input Dimension Reduction for Gaussian Process Modeling
2026-07-23T07:24:04Z•c074ff47524ccf4e3fcf9d5393186f4ff5bf55804d1c2c9ee5088b10a29347e8
DKMDLoRARELTA-SGLDStatLoRAannealingappend-only-attacksbayesiancausal-inferencecovariate-shiftdata-poisoningdeep-gaussian-processdistribution-shiftflowsgaussian-processhamiltonian-monte-carlolow-rank-adaptationmachine-learningmodel-fine-tuningmolecular-designpoisoning-auditsde-trainingsghmc-sgldstiefel-manifoldtotal-influencetraining-stability
What happened
Collection of recent machine-learning papers spanning Bayesian dimension reduction for Gaussian processes (Stiefel-manifold priors, HMC geodesic flow), DECAF annealing flows for Boltzmann-expected 3D molecular design, RELTA-SGLD (a localized taming scheme improving stability for superlinear stochastic-gradient Langevin dynamics), optimal online predictor recalibration, and several applied/statistical contributions (fast DKMD signed statistic for univariate distribution shifts, an R package for NMF, SPDNN for covariate shift, and adaptive Bayesian online aggregation). Security-relevant items: “
Why it matters
A reviewed impact interpretation has not been published for this record.
Evidence and limitations
- Source ID
- arxiv_stat_ml
- Record identifier
- c074ff47524ccf4e3fcf9d5393186f4ff5bf55804d1c2c9ee5088b10a29347e8
- Enrichment time
- 2026-07-23T07:24:04Z
- AI-assisted enrichment
- Yes
This record may overlap with other records. Its enrichment can be incomplete or wrong, and machine assistance was used. Validate consequential decisions against the linked source and your own environment.