AI Paper Atlas

Research map / Optimization / Optimizers

Multivariate Triangular Quantile Maps for Novelty Detection

Jingjing Wang, Sun Sun, Yaoliang Yu

NeurIPS 2019 · Optimization › Optimizers · approach: Surrogate losses and robust solvers · 11 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
train autoencoder and flow jointly with MGDA instead of freezing autoencoder after pretraining (Fix-) (ablation)Ours-NLL · MNISTAUROC: 0.9699 -> 0.9728Improved
joint MGDA training instead of pretrain then joint with fixed lambda 0.5 (ablation)Ours-NLL · MNISTAUROC: 0.9703 -> 0.9728Improved
joint MGDA training instead of fixed two-stage (Fix-) (ablation)Ours-TQM2 · MNISTAUROC: 0.9634 -> 0.9699Improved
MGDA-optimized lambda instead of fixed lambda 0.5 (ablation)Ours-NLL · MNISTAUROC: 0.9537 -> 0.9728Improved
MGDA-optimized lambda instead of best fixed lambda 0.99 (ablation)Ours-NLL · MNISTAUROC: 0.9729 -> 0.9728No clear effect
use density (NLL) scoring instead of 1-norm quantile (TQM1) scoring (ablation)Ours · MNISTAUROC: 0.9666 -> 0.9728Improved
use quantile infinity-norm scoring instead of density scoring (ablation)Ours · MNISTAUROC: 0.9728 -> 0.9531Hurt
add flow-based density/quantile scoring on latent instead of reconstruction error only (ablation)Ours · KDDCUPF1: 0.6296 -> 0.9622Improved
use TQM-infinity scoring instead of NLL scoring (ablation)Ours · ThyroidF1: 0.7312 -> 0.7527Improved
Ours-NLL vs GT, GPND on MNISTOurs · MNISTAUROCMixed

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