AI Paper Atlas

Research map / Continual learning / Class-incremental

On the Soft-Subnetwork for Few-Shot Class Incremental Learning

Haeyong Kang, Jaehong Yoon, Sultan Rizky Hikmawan Madjid, Sung Ju Hwang, Chang D. Yoo

ICLR 2023 · Continual learning › Class-incremental · approach: Few-shot class-incremental prototypes · 12 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 a soft minor subnetwork instead of a hard subnetwork at c=50% (ablation)SoftNet · CIFAR-100 session 9classification accuracy: 54.32 -> 55.33Improved
use a soft minor subnetwork instead of a hard subnetwork at c=80% (ablation)SoftNet · CIFAR-100 session 9classification accuracy: 54.31 -> 54.94Improved
use a soft minor subnetwork instead of a hard subnetwork at c=90% (ablation)SoftNet · CIFAR-100 session 9classification accuracy: 53.92 -> 54.76Improved
use a soft minor subnetwork instead of a hard subnetwork at c=80% (ablation)SoftNet · miniImageNet session 9classification accuracy: 52.74 -> 54.34Improved
use a soft minor subnetwork instead of a hard subnetwork at c=87% (ablation)SoftNet · miniImageNet session 9classification accuracy: 53.47 -> 54.68Improved
use a soft minor subnetwork instead of a hard subnetwork at c=90% (ablation)SoftNet · miniImageNet session 9classification accuracy: 53.07 -> 54.14Improved
use a soft minor subnetwork instead of a hard subnetwork at c=88% (ablation)SoftNet · CUB-200-2011 session 11classification accuracy: 55.20 -> 56.64Improved
use a soft minor subnetwork instead of a hard subnetwork at c=90% (ablation)SoftNet · CUB-200-2011 session 11classification accuracy: 55.30 -> 56.75Improved
use a soft minor subnetwork instead of a hard subnetwork at c=93% (ablation)SoftNet · CUB-200-2011 session 11classification accuracy: 55.18 -> 56.28Improved
use SoftNet rather than ALICEResNet18 FSCIL · CIFAR-100 session 9classification accuracy: 54.10 -> 55.33Improved
use SoftNet rather than ALICEResNet18 FSCIL · miniImageNet session 9classification accuracy: 55.70 -> 54.68Hurt
use SoftNet rather than ALICEResNet18 FSCIL · CUB-200-2011 session 11classification accuracy: 60.10 -> 56.75Hurt

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