AI Paper Atlas

Research map / Time series / Anomaly detection

PGRF-Net: A Prototype-Guided Relational Fusion Network for Diagnostic Multivariate Time-Series Anomaly Detection

Jahoon Jeong, Hyunsoo Yoon

ICLR 2026 · Time series › Anomaly detection · approach: Multivariate anomaly detection

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
add frequency decomposition to the encoder (ablation)PGRF-Net (Conformer) · Five MTSAD datasets (avg)Avg F1: 95.12 -> 97.41Improved
use Conformer instead of Transformer/PatchTST/MLP-Mixer encoder (ablation)PGRF-Net with FD · Five MTSAD datasets (avg)Avg F1: 96.82 -> 97.41 (vs PatchTST)Improved
add Stage-2 gated fusion over static equal-weight fusion (ablation)PGRF-Net · Five MTSAD datasets (avg)Avg F1: 96.82 -> 97.41Improved
add structural, contextual and spike evidence to predictive score (ablation)PGRF-Net · Five MTSAD datasets (avg)Avg F1: 92.51 -> 96.82Improved
include structural evidence channel in static fusion (ablation)PGRF-Net static fusion · Five MTSAD datasets (avg)Avg F1: 94.13 -> 96.82Improved
include contextual evidence channel in static fusion (ablation)PGRF-Net static fusion · Five MTSAD datasets (avg)Avg F1: 94.58 -> 96.82Improved
include spike evidence channel in static fusion (ablation)PGRF-Net static fusion · Five MTSAD datasets (avg)Avg F1: 95.34 -> 96.82Improved
use learned weighted fusion rather than max pooling of four scores (ablation)PGRF-Net · Five MTSAD datasets (avg)Avg F1: 92.32 -> 97.41Improved
use two-stage training instead of end-to-end joint training (ablation)PGRF-Net · Five MTSAD datasets (avg)Avg F1: 95.12 -> 97.41Improved
increase batch size from 128 to 512 (ablation)PGRF-Net · Five MTSAD datasets (avg)F1: 97.41 -> 92.63Hurt
increase acyclicity regularization from 1e-4 to 1e-2 (ablation)PGRF-Net · Five MTSAD datasets (avg)F1: 97.41 -> 89.88Hurt
PGRF-Net versus H-PAD on point-wise non-adjusted F1PGRF-Net · Five MTSAD datasets (point-wise non-adjusted)Avg F1: 46.59 vs 46.76 (InterFusion)No clear effect

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.

Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.

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