AI Paper Atlas

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

Wisdom of the Ensemble: Improving Consistency of Deep Learning Models

Lijing Wang, Dipanjan Ghosh, Maria Gonzalez Diaz, Ahmed Farahat, Mahbubul Alam, Chetan Gupta, Jiangzhuo Chen, Madhav Marathe

NeurIPS 2020 · Learning theory and algorithms › Prediction and decision losses · approach: Data attribution and recourse · 2 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
Combine accuracy-selected snapshot learners across resampled training runs using dynamic pruningDynSnap-cyc · CIFAR100ACC-CON (%)Improved
Use DynSnap-cyc instead of unpruned Snapshot (β=1)ResNet20 · CIFAR10CON (%): 89.88 -> 93.04Improved
Use DynSnap-cyc instead of unpruned Snapshot (β=1)ResNet56 · CIFAR100ACC-CON (%): 66.14 -> 71.37Improved
Use DynSnap-cyc instead of unpruned Snapshot (β=1)fastText · YAHOO!AnswersCON (%): 91.87 -> 92.88Improved
Replace SingleBase with an ensemble of component learnersdeep classifiers · CIFAR10, CIFAR100, and YAHOO!AnswersCON (%): 4.5%-8.2%, 9.8%-19.3%, and 3.4%-4.5%Improved
Replace SingleBase with MCDropoutbase classifiers · CIFAR10, CIFAR100, and YAHOO!AnswersACC (%)Hurt
Increase ensemble size m (ablation)DynSnap-cyc with AVG · CIFAR100performanceImproved
Increase pruning factor β, reducing pruning and admitting relatively poorer learners (ablation)DynSnap-cyc with AVG · CIFAR100performanceHurt
Increase snapshot count N (ablation)DynSnap-cyc with AVG · CIFAR100performanceImproved
Increase snapshot count beyond N=20 (ablation)DynSnap-cyc with AVG · CIFAR100performance: N=20No clear effect
Vary snapshot window size T/N (ablation)DynSnap-cyc with AVG · CIFAR100performanceNo clear effect
Remove the learning-rate schedule (ablation)DynSnap-step, top-N snapshot · CIFAR100performanceHurt

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