OptIDC: Optimal Transport Unifies Sample, Feature and Embedding Selection for CTR Prediction
_reachsumit · x · 2026-10-28
A new paper proposes OptIDC, a unified framework that jointly learns sample, feature, and embedding-unit level configurations for deep CTR prediction, where prior work optimized each input-side selection mechanism independently. It introduces a selection-weighted Optimal Transport mechanism aligning coarse- and fine-grained selections to ease cross-level conflicts while preserving each level's semantics. The framework is model-agnostic and plugs into existing feature interaction architectures.
More from Research
- Frontier models hit 73.3% on SpatialBench 10 months after release, near 75% ceiling — kenbwork · 2026-10-28
- Minimal PyTorch project implements transformer autograd by hand for learning custom backward passes — ifaposto · 2026-10-28
- Ben Goertzel's toy simulations probe the geometry of superintelligence attractor basins — bengoertzel · 2026-10-28
- ReGE: turning recorded agent execution experience into synthesized guidance for self-improvement — TIGER-Lab · 2026-10-28
- Statistician: the Brazil event wasn't 'impossible' — the distribution was wrong — doodlestein · 2026-10-28
- NVIDIA PhysicsNeMo tutorial: AI surrogate models for vehicle crash analysis — NVIDIA Developer · 2026-10-28