/Rotated_MNIST_Continual_Learning

Continual learning strategies(EWC, GEM) for rotated MNIST dataset

Primary LanguageJupyter Notebook

Rotated_MNIST_Continual_Learning

Continual learning strategies(EWC, GEM) for rotated MNIST dataset

Group Memeber:

Ruinan Zhang rz1109@nyu.edu Manlan Li ml6589@nyu.edu

Project Description

In this projct, our group exlpored the rotated MNIST dataset with two continual learning strategies:

Dataset

Training Dataset: 60000 * 28 * 28

Test Dataset: 10000 * 28 * 28

Rotation: Randomized rotation

MNIST VS Rotated MNIST

Rotated MNIST

Result

  • (1) EWC run on total 10 tasks with final acc 85.78%:

EWC res

  • (2) GEM run on total 20 tasks with final acc 90.21%

GEM res

GEM res2

Besides accuracy, there are two additional metrics used to assess the ability of the algorithm to transfer knowledge - Backward transfer (BWT) and Forward transfer (FWT)

BWT: the influence that learning a task t has on the performance on a previous task k ≺ t

FWT: the influence that learning a task t has on the performance on a future task k ≻ t

GEM res2