AI Paper Atlas

Research map / Learning theory and algorithms / Combinatorial and search optimization

Finer Metagenomic Reconstruction via Biodiversity Optimization

Simon Foucart, David Koslicki

NeurIPS 2020 · Learning theory and algorithms › Combinatorial and search optimization · approach: Decision-focused optimization

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
use phylogenetic similarity matrix instead of identity or random similarity matrix (ablation)MinDiv via fmincon · 64 x 192 k-mer matrix, k=3, 200 replicatespercent recovered vectorsImproved
use diversity minimization with phylogenetic matrix instead of QuikrMinDiv · 64 x 192 k-mer matrixpercent recovered vectorsImproved
use phylogenetic similarity matrix at realistic scaleMinDiv · larger problemscomputationHurt
use taxonomic block-diagonal diversity (IRWLP) instead of QuikrIRWLP · 256 x 768 k-mer matrix, genus-level recovery, support 70 to 150percent recovered vectorsImproved
use IRWLP instead of Quikr when recovery succeedsIRWLP · 256 x 768 k-mer matrixexecution time (s)Improved
use IRWLP instead of Quikr when recovery mostly failsIRWLP · 256 x 768 k-mer matrix, large supportexecution time (s)Hurt
compare IRWLP with a feasibility-only procedure (ablation)IRWLP · 256 x 768 k-mer matrixpercent recovered vectorsImproved
use co-occurrence MinDivLP (h=6, 13) instead of standard l1-minimizationMinDivLP · 256 x 768 k-mer matrix, support 25 to 70percent recovered vectorsImproved
use MinDivLP with h > k instead of weighted approach (h = k = 4) (ablation)MinDivLP · 256 x 768 k-mer matrixpercent recovered vectorsImproved
increase h from 4 to 6 to 13 (ablation)MinDivLP · 256 x 768 k-mer matrixpercent recovered vectorsImproved
use MinDivLP instead of QuikrMinDivLP · 256 x 768 k-mer matrixpercent recovered vectorsImproved
detect failure a posteriori with residual of multi-k-mer constraintsMinDivLP with k=4, h=13 · 256 x 768 k-mer matrixfailure detectionImproved

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