Separation Capacity of Scattering Networks
2026-07-01T07:23:52Z•2e16984596082558eec3165106e84fe488edba1df6d5045b1841e69362e8f272
algorithmsarxivbenchmarkscausal-inferenceconformal-predictiondeep-learninggaussian-processesmachine-learningmissing-dataoptimizationspatio-temporalstatisticstheoryuncertainty-quantification
What happened
This feed contains a set of new/stat_ml arXiv papers (July 1, 2026) covering theoretical and applied advances across machine learning and statistics: (1) a Cover-theory-based analysis of separation capacity for scattering networks and guidance for their design; (2) a neural-network framework for online dynamic prediction of alternating recurrent events using inverse-probability-weighted pseudo-observations; (3) sharp convergence results for SGD in the stochastic edge-of-stability regime for cross-entropy loss (linear and two-layer nets) showing stochastic self-stabilization with large steps; (
Why it matters
A reviewed impact interpretation has not been published for this record.
Evidence and limitations
- Source ID
- arxiv_stat_ml
- Record identifier
- 2e16984596082558eec3165106e84fe488edba1df6d5045b1841e69362e8f272
- Enrichment time
- 2026-07-01T07:23:52Z
- 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.