Channel-wise-Lightweight-Reprogramming

[ICCV 2023] CLR: Channel-wise Lightweight Reprogramming for Continual Learning

Project Page(pending) | Video(Youtube) | Video(bilibili) | Paper

CLR: Channel-wise Lightweight Reprogramming for Continual Learning
Yunhao Ge, Yuecheng Li, Shuo Ni, Jiaping Zhao, Ming-Hsuan Yang, Laurent Itti
ICCV

Figure: Equipped with a task-agnostic immutable CNN model, our approach "reprogram" the CNN layers to each new task with lightweight task-specific parameters (less than 0.6% of the original model) to learn sequences of disjoint tasks, assuming data from previous tasks is no longer available while learning new tasks.

We propose a Channel-wise Lightweight Reprogramming (CLR) approach that helps convolutional neural networks (CNNs) overcome catastrophic forgetting during continual learning. We show that a CNN model trained on an old task (or self-supervised proxy task) could be "reprogrammed" to solve a new task by using our proposed lightweight (very cheap) reprogram- ming parameter. With the help of CLR, we have a better stability-plasticity trade-off to solve continual learning problems: To maintain stability and retain previous task ability, we use a common task-agnostic immutable part as the shared “anchor” parameter set. We then add task-specific lightweight reprogramming parameters to reinterpret the outputs of the immutable parts, to enable plasticity and integrate new knowledge

Figure: Proposed continual learning model with channel-wise lightweight reprogramming (CLR) layers. All gray blocks are fixed parameters. (top) General network architecture. (bottom) Details of CLR reprogramming layer: for each channel k ∈ [1..c] of an original w × h × c feature map (blue), a 3x3 kernel is learned to reprogram the feature towards the new task (green), without modifying the original conv parameters (grey).

Getting starged

Installation

git clone https://github.com/gyhandy/Channel-wise-Lightweight-Reprogramming.git

Download file

Single file downloading option (~1.6T)

wget http://ilab.usc.edu/andy/skill-dataset/skill/SKILL-Dataset-backend.zip
unzip SKILL-Dataset-backend.zip

Multiple files downloading option(~600G per file)

wget http://ilab.usc.edu/andy/skill-dataset/separate-v/skill-dataset.z01
wget http://ilab.usc.edu/andy/skill-dataset/separate-v/skill-dataset.z02
wget http://ilab.usc.edu/andy/skill-dataset/separate-v/skill-dataset.zip
unzip skill-dataset.zip

General directory structure

• dataset/ contains code for declare train_datasets, val_datasets, train_loaders, and val_loaders for the DCT dataset, each is a list of 107 datasets contains in the DCT. You can also define your own train_datasets, val_datasets, train_loaders, and val_loaders for your own datasets. Place to change is commented in the main code

• network/ contains models and specific customized layer used in the experiment

• main.py The main code needs to be run

• stat.csv The stats and subset id of the 53-task we used from DCT dataset.

Usage

argument

• --result the result path will store all the logs

• --weight the weight path which store the weight of the classifiers

• --data the path to store the data, this should be where you unzip before

• --method Ghost_reduced_v1 refers to the main method in the paper while Ghost provides a slightly expensive version in memory cost, but come with a slight accuracy improvment

Sample run

• To run the standard version of CLR

python main.py --data <Folder where you unzip>

The result can be found in Figure 11 under column "CLR"

• To run a slight expanded version of CLR

python main.py --data <Folder where you unzip> --method Ghost

The result can be found in Figure 11 under column "CLR-Full"