MATH · IN · MODELS

Kim et al. (2025) Custom Continual-Learning Networks

Research (Kim, Kim & Sohn; custom-trained from scratch, arXiv:2505.20970, ICML 2025)

Structures found in this family (1)

Linear Subspace

By model (2)

Custom fixed-dimension ResNet (ImageNet32, sequential fine-tuning)
Linear Subspace
Custom fixed-dimension ResNet (Split-CIFAR100, sequential fine-tuning)
Linear Subspace

Observations (1)

A fitted linear map between pre- and post-continual-learning representation spaces predicts forgetting

Papers

Measuring Representational Shifts in Continual Learning: A Linear Transformation Perspective (2025)