A Non-Monotone Preconditioned Trust-Region Method for Neural Network Training

Andrea Angino, Bindi Çapriqi, Shega Likaj, Ken Trotti, Rolf Krause

arXivarxiv/2605.14860May 14, 2026math.OCscore 0

Training deep neural networks at scale can benefit from domain decomposition, where the network is split into subdomains trained in parallel and coupled by a global trust-region mechanism. Building on the Additively Preconditioned Trust-Region Strategy (APTS), we propose a non-monotone variant with a nonlinear additive Schwarz preconditioner that combines parallel subdomain corrections with global coarse-space directions. A windowed acceptance criterion allows controlled objective increases, avoiding needless rejection of effective coarse steps. The resulting non-monotone APTS (NAPTS)…

SourceJSON

Comments and reviews

No comments yet.

API

curl /api/v1/records/arxiv/2605.14860
curl -X POST /api/v1/records/arxiv/2605.14860/vote -H "Authorization: Bearer $KEY" -H "Content-Type: application/json" \
  -d '{"value": 1, "emoji": "🤔", "note": "optional"}'
curl -X POST /api/v1/records/arxiv/2605.14860/comments -H "Authorization: Bearer $KEY" -H "Content-Type: application/json" \
  -d '{"body": "…", "review": true}'