AI Paper Atlas

Research map / Architectures and efficiency / RNNs and normalization

Deep Elastic Networks With Model Selection for Multi-Task Learning

Chanho Ahn, Eunwoo Kim, Songhwai Oh

ICCV 2019 · Architectures and efficiency › RNNs and normalization · approach: Differentiable optimization layers · 49 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
use DEN with rho=0.1 instead of separate baseline models on three tasksResNet-42 · STL-10 (M1)accuracy (%): 76.25 -> 87.58Improved
use DEN (rho=0.1) instead of baselineResNet-42 · Tiny-ImageNet (M1)accuracy (%): 57.22 -> 60.21Improved
use DEN (rho=0.1) with one shared network instead of three baselinesResNet-42 · M1 average of three tasksaccuracy (%): 69.51 -> 74.30Improved
use DEN (rho=0.1) instead of baselineWRN-32-4 · M1 average of three tasksaccuracy (%): 71.26 -> 73.82Improved
raise sparsity weight rho from 0.1 to 1 (ablation)DEN · M1 average (ResNet-42)accuracy (%): 74.30 -> 71.65Hurt
raise rho from 0.1 to 1 (ablation)DEN · M1 FLOPs (ResNet-42)FLOPs: 1.61G -> 1.35GImproved
use DEN (rho=0.1) instead of RoutingMini-ImageNet MTL · Mini-ImageNet (M2, 224x224)accuracy (%): 58.97 -> 65.23Improved
use DEN (rho=1) instead of baseline on coarse labelsWRN-32-4 · CIFAR-100 coarse (M3)accuracy (%): 83.53 -> 84.87Improved
use DEN (rho=1) instead of baseline on fine labelsWRN-32-4 · CIFAR-100 fine (M3)accuracy (%): 76.32 -> 75.93No clear effect
add the learned selector instead of a random selector (ablation)DEN ResNeXt-29 (8x64d) · CIFAR-10 compression (C1)accuracy (%): 90.55 -> 94.61Improved
add the learned selector instead of a random selector (ablation)DEN ResNeXt-29 (4x64d) · CIFAR-100 compression (C1)accuracy (%): 65.44 -> 77.58Improved
increase levels of hierarchy per block up to four, then to five (ablation)DEN WRN-32-4 · CIFAR-100accuracyMixed

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