AI Paper Atlas

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

Efficient Data Subset Selection to Generalize Training Across Models: Transductive and Inductive Networks

Eeshaan Jain, Tushar Nandy, Gaurav Aggarwal, Ashish V. Tendulkar, Rishabh K Iyer, Abir De

NeurIPS 2023 · Learning theory and algorithms › Prediction and decision losses · approach: Active learning and subset selection · 3 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
replace uncertainty sampler with SubSelNet inductive sampler (ablation)Attn. g_beta · CIFAR10 b=0.03|D|RAR: 0.794 -> 0.429Improved
replace loss sampler with SubSelNet inductive sampler (ablation)Attn. g_beta · CIFAR10 b=0.1|D|RAR: 0.407 -> 0.260Improved
replace feedforward approximator with attention approximator (ablation)Inductive sampler · CIFAR10 b=0.1|D|RAR: 0.397 -> 0.260Improved
replace LSTM approximator with attention approximator (ablation)Inductive sampler · CIFAR10 b=0.05|D|RAR: 0.412 -> 0.310Improved
use attention approximator instead of feedforward (ablation)uncertainty sampler · CIFAR10 b=0.1|D|RAR: 0.547 -> 0.679Hurt
use attention approximator instead of LSTM/feedforward (ablation)model approximator · CIFAR10KL divergence: 0.171 -> 0.089Improved
use g_beta directly to predict accuracy instead of subset training pipeline (ablation)SubSelNet · CIFAR10 b=90%RAR difference: -0.487Hurt
use Transductive instead of Inductive SubSelNetSubSelNet · FMNIST, CIFAR10, CIFAR100RAR vs speedup trade-offImproved
tune B in Hybrid-SubSelNet (B from 25K to 50K)Hybrid-SubSelNet · CIFAR100trade-offMixed
use SubSelNet (transductive) subsets in NAS instead of randomDARTS NAS · CIFAR10 b=10%test error (%): 3.02 -> 2.82Improved
use SubSelNet subsets in HPO instead of randomHPO · CIFAR10 b=5%test error (%): 5.4 -> 4.11Improved
use SubSelNet subsets in HPO instead of AUTOMATAHPO · CIFAR10 b=10%test error (%): 3.39 -> 2.70Improved

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.

Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.

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