AI Paper Atlas

Research map / Learning theory and algorithms / Prediction and decision losses

Scalable Air-Quality Sensor Placement via Gradient-Based Mutual Information Maximization

Zeel B Patel, Vinayak Rana, Nipun Batra

AAAI 2026 · Learning theory and algorithms › Prediction and decision losses · approach: Adaptive testing and experiments

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
jointly optimize sensor coordinates by gradient descent instead of greedy MaxVar (ablation)sensor placement with TNP-D surrogate · India PM2.5 test set, 100 added stationsRMSE: 7.4 -> 7.1Improved
replace greedy MaxVar with GD-MI (ablation)sensor placement, Madhya Pradesh · Madhya Pradesh k=9 (validation)RMSE: 6.3 -> 5.8Improved
use greedy MI instead of GD-MI (ablation)sensor placement, Madhya Pradesh · Madhya Pradesh k=9 (validation)RMSE: 5.8 -> 5.6Improved
increase budget of added stations (ablation)GD-MI versus MaxVar · India PM2.5 test setRMSE gapImproved
use TNP-D surrogate instead of Gaussian Process (ablation)PM2.5 surrogate model · WUSTL PM2.5 validationRMSE: 5.16 -> 4.90Improved
use attentive or convolutional NP instead of basic CNP (ablation)Neural Process surrogate · WUSTL PM2.5 validationRMSE: 11.46 -> 5.28 (ConvCNP)Improved
use ConvGNP with full covariance instead of ConvCNP (ablation)Neural Process surrogate · WUSTL PM2.5 validationRMSE: 5.28 -> 5.31No clear effect
add out-of-region penalty to keep sensors on valid landGD-MI · India PM2.5placement validityImproved

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