AI Paper Atlas

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

Adaptive stimulus selection for optimizing neural population responses

Benjamin Cowley, Ryan Williamson, Katerina Clemens, Matthew Smith, Byron M. Yu

NeurIPS 2017 · 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
use Adept instead of random image selectionAdept · CNN testbedmean neuron response: 15.5% largerImproved
use Adept instead of random image selectionAdept · CNN testbedresponse scatter (PCA eigenvalue ratio)Improved
population objective (Adept) instead of single-neuron objectives (Adept-1, Adept-50, genetic-50) (ablation)Adept · CNN testbedmean response and eigenvalue ratioImproved
keep only norm term in objective (Adept-norm) (ablation)Adept · CNN testbedmean response: p = 0.65No clear effect
keep only norm term in objective (Adept-norm) (ablation)Adept · CNN testbedresponse scatter (eigenvalue ratio): p < 0.005Hurt
keep only average-distance term (Adept-avgdist) (ablation)Adept · CNN testbedmean response: p < 10Hurt
keep only average-distance term (Adept-avgdist) (ablation)Adept · CNN testbedresponse scatter (eigenvalue ratio): p = 0.62No clear effect
use embedding layer other than middle layer of GoogLeNet (ablation)Adept feature embeddings · CNN testbedcorrelation predicted vs actual objective: 0.6No clear effect
use Adept instead of random selection in macaque V4Adept · V4 recordings, 7 sessionsmean response: 19.5% higherImproved
use Adept in V4 for response scatterAdept · V4 recordingseigenvalue ratio by PC dimensionMixed
vary CNN layer used for V4 feature embeddings (deeper 5-10 vs 1-4) (ablation)Adept feature embeddings · V4 recordingscorrelation predicted vs actual objectiveImproved
replace CNN embeddings with V4 responses from another day (ablation)Adept feature embeddings · V4 recordingscorrelation predicted vs actual objective: 0.25 -> 0.5Improved

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