AI Paper Atlas

Research map / Optimization / Nonconvex and smooth optimization

A Game of Bundle Adjustment - Learning Efficient Convergence

Amir Belder, Refael Vivanti, Ayellet Tal

ICCV 2023 · Optimization › Nonconvex and smooth optimization · approach: Equilibrium learning in zero-sum games · 6 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
learn lambda holistically with RL (SAC) instead of non-holistic Zero-Net supervised scheme (ablation)Ours + classic · BALiterations to converge: 8 -> 4Improved
use state size 5 instead of 1 (consecutive estimation errors) (ablation)Ours + classic · BALiterations to converge: 7.5 -> 4Improved
increase state size from 5 to 20 (ablation)Ours + classic · BALiterations to converge: 4 -> 5Hurt
use estimation error as state and time as reward instead of reversed roles (ablation)Ours + classic · BALiterations / success rate: 10 -> 4 iterations; 70% -> 100% successImproved
use iteration-duration reward instead of constant negative reward (ablation)Ours + classic · BALiterations to converge: 5 -> 4Improved
use iteration-duration reward instead of reward reduction (ablation)Ours + classic · BALiterations to converge: 4.2 -> 4Improved
use learned dynamic lambda instead of a fixed lambda schedule (ablation)Ours + classic · BALiterations to converge: 12 -> 4Improved
train the agent on small synthetic random-point scenes instead of KITTI/BAL (ablation)Ours + classic · KITTI and BALiterations to converge: 14 -> 17 (KITTI); 4 -> 5 (BAL)No clear effect
apply learned damping on very small BA problems (5 points, 5 poses)Ours + classic · synthetic small problemsiterations / time: 5 -> 4 iterations but about 0.5 s slower overallHurt
use the learned damping agent with a high tolerance (0.1)Ours + classic · BAL performance profilesolve timeNo clear effect

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy