AI Paper Atlas

Research map / Architectures and efficiency / Neural architecture search

Learning from Mistakes – a Framework for Neural Architecture Search

Bhanu Garg, Li Zhang, Pradyumna Sridhara, Ramtin Hosseini, Eric Xing, Pengtao Xie

AAAI 2022 · Architectures and efficiency › Neural architecture search · approach: AutoML pipeline search · 6 citations (OpenAlex)

Read the paper

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
Apply LFM to DARTS-2ndDARTS-2nd · CIFAR-100test error (%): 20.58\pm0.44 -> 17.65\pm0.45Improved
Apply LFM to PDARTS with a ResNet-18 encoderPDARTS · CIFAR-100test error (%): 17.49 -> 16.44\pm0.11Improved
Use a ResNet-34 encoder instead of ResNet-18 (ablation)LFM-PDARTS · CIFAR-100test error (%): 0.75%Improved
Apply LFM to DARTS-2ndDARTS-2nd · CIFAR-10test error (%): 2.76\pm0.09 -> 2.70\pm0.06Improved
Apply LFM to PDARTSPDARTS · CIFAR-10test error (%): 2.50 -> 2.46\pm0.04Improved
Apply LFM to DARTS-2nd architecture searched on CIFAR-10DARTS2nd-CIFAR10 · ImageNettop-1 error (%): 26.7 -> 25.12Improved
Apply LFM to PDARTS architecture searched on CIFAR-10PDARTS (CIFAR10) · ImageNettop-5 error (%): 7.4 -> 6.85Improved
Use an encoder trained only on CIFAR-100 rather than ImageNet-pretrained (ablation)LFM+DARTS · CIFAR-100test error (%): 17.65\pm0.45 -> 17.82\pm0.39No clear effect
Use an encoder trained only on CIFAR-100 rather than ImageNet-pretrained (ablation)LFM+PDARTS · CIFAR-100test error (%): 16.44\pm0.11 -> 16.51\pm0.13No clear effect
Add parameter sharing between learner models and use W1 as the encoder (ablation)LFM+DARTS · CIFAR-100test error (%): 17.65\pm0.45 -> 18.77\pm0.31Hurt
Add parameter sharing between learner models and use W1 as the encoder (ablation)LFM+PDARTS · CIFAR-100test error (%): 16.44\pm0.11 -> 16.83\pm0.08Hurt
Use dot-product attention rather than cosine similarity or L2 distance for visual similarity (ablation)LFM · CIFAR-100test errorImproved

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