Elastic weight consolidation technique for incremental learning.

Related tags

Deep Learning neural-network deepmind incremental-learning elastic-weight-consolidation
Overview

Overcoming-Catastrophic-forgetting-in-Neural-Networks

Elastic weight consolidation technique for incremental learning.

About

Use this API if you dont want your neural network to forget previously learnt tasks while doing transfer learning or domain adaption!

Results

The experiment is done as follow:

  1. Train a 2 layer feed forward neural network on MNIST for 4 epochs
  2. Train the same network later on Fashion-MNIST for 4 epochs This is done once with EWC and then without EWC and results are calculated on test data for both data on same model. Constant learning rate of 1e-4 is used throughout with Adam Optimizer. Importance multiplier is kept at 10e5 and sampling is done with half data before moving to next dataset
EWC MNIST Fashion-MNIST
Yes 70.27 81.88
No 48.43 86.69

Usage

from elastic_weight_consolidation import ElasticWeightConsolidation
# Build a neural network of your choice and pytorch dataset for it
# Define a criterion class for new task and pass it as shown below
ewc = ElasticWeightConsolidation(model, crit, lr=0.01, weight=0.1)
# Training procedure
for input, target in dataloader:
  ewc.forward_backward_update(input, target)
ewc.register_ewc_params(dataset, batch_size, num_batches_to_run_for_sampling)
# Repeat this for each new task and it's corresponding dataset

Reference

Paper

You might also like...
This is the formal code implementation of the CVPR 2022 paper 'Federated Class Incremental Learning'.
This is the formal code implementation of the CVPR 2022 paper 'Federated Class Incremental Learning'.

Official Pytorch Implementation for GLFC [CVPR-2022] Federated Class-Incremental Learning This is the official implementation code of our paper "Feder

Race Wang 57 Dec 27, 2022
Official Pytorch implementation of Online Continual Learning on Class Incremental Blurry Task Configuration with Anytime Inference (ICLR 2022)
Official Pytorch implementation of Online Continual Learning on Class Incremental Blurry Task Configuration with Anytime Inference (ICLR 2022)

The Official Implementation of CLIB (Continual Learning for i-Blurry) Online Continual Learning on Class Incremental Blurry Task Configuration with An

NAVER AI 34 Oct 26, 2022
Sound and Cost-effective Fuzzing of Stripped Binaries by Incremental and Stochastic Rewriting
Sound and Cost-effective Fuzzing of Stripped Binaries by Incremental and Stochastic Rewriting

StochFuzz: A New Solution for Binary-only Fuzzing StochFuzz is a (probabilistically) sound and cost-effective fuzzing technique for stripped binaries.

Zhuo Zhang 164 Dec 5, 2022
🔥 TensorFlow Code for technical report:
🔥 TensorFlow Code for technical report: "YOLOv3: An Incremental Improvement"

🆕 Are you looking for a new YOLOv3 implemented by TF2.0 ? If you hate the fucking tensorflow1.x very much, no worries! I have implemented a new YOLOv

null 3.6k Dec 26, 2022
Incremental Transformer Structure Enhanced Image Inpainting with Masking Positional Encoding (CVPR2022)
Incremental Transformer Structure Enhanced Image Inpainting with Masking Positional Encoding (CVPR2022)

Incremental Transformer Structure Enhanced Image Inpainting with Masking Positional Encoding by Qiaole Dong*, Chenjie Cao*, Yanwei Fu Paper and Supple

Qiaole Dong 190 Dec 27, 2022
An open source machine learning library for performing regression tasks using RVM technique.

Introduction neonrvm is an open source machine learning library for performing regression tasks using RVM technique. It is written in C programming la

Siavash Eliasi 33 May 31, 2022
code for paper "Not All Unlabeled Data are Equal: Learning to Weight Data in Semi-supervised Learning" by Zhongzheng Ren*, Raymond A. Yeh*, Alexander G. Schwing.

Not All Unlabeled Data are Equal: Learning to Weight Data in Semi-supervised Learning Overview This code is for paper: Not All Unlabeled Data are Equa

Jason Ren 22 Nov 23, 2022
CoReNet is a technique for joint multi-object 3D reconstruction from a single RGB image.
CoReNet is a technique for joint multi-object 3D reconstruction from a single RGB image.

CoReNet CoReNet is a technique for joint multi-object 3D reconstruction from a single RGB image. It produces coherent reconstructions, where all objec

Google Research 80 Dec 25, 2022
Pytorch implementation of AngularGrad: A New Optimization Technique for Angular Convergence of Convolutional Neural Networks
Pytorch implementation of AngularGrad: A New Optimization Technique for Angular Convergence of Convolutional Neural Networks

AngularGrad Optimizer This repository contains the oficial implementation for AngularGrad: A New Optimization Technique for Angular Convergence of Con

mario 124 Sep 16, 2022
Comments
  • this ewc implementation CODE has theoretical ERROR which prevent ewc to work properly

    this ewc implementation CODE has theoretical ERROR which prevent ewc to work properly

    at line 31 of elastic_weight_consolidation.py it calculates mean of log_likelihoods so grad_log_liklihood will contain mean of gradients of log_likelihoods and then at line 35 it squares this mean of gradients of log_likelihoods. this is WRONG because diagonal element of Fisher matrix is sum of squared gradients of log_liklihoods but not squared sum of gradients of log_liklihoods. so for each input the separate gradient of log_likelihood must be calculated, then each gradient must be squared and then mean of these squares must be calculated/

    opened by aakutalev 0
  • use torch.gather instead of direct indexing

    use torch.gather instead of direct indexing

    Instead of this line:

    log_liklihoods.append(output[:, target])

    have this line:

    log_liklihoods.append(torch.gather(output, dim=1, index=target.unsqueeze(-1)))

    Why?

    Assume our output is 100x4 which means batch size is 100 and we have 4 classes. Target is a (100,) vector of classes, by indexing output[:, target] we will create a 100x100 matrix, instead of gathering the loglikelihoods 100x1 that we desire.

    The torch.gather function does this propoerly.

    opened by afshinrahimi 0
  • Fisher Update causing errors

    Fisher Update causing errors

    I am trying to run EWC on my dataset with resnet50 model. While updating the fisher matrix using your function, My code says Cuda out of memory due to "log_liklihoods.append(output[:, target])" in the code. I read this "https://stackoverflow.com/questions/59805901/unable-to-allocate-gpu-memory-when-there-is-enough-of-cached-memory" and figured out the problem using 'detach()'. After doing detach etc, I get an error: RuntimeError: One of the differentiated Tensors appears to not have been used in the graph. Set allow_unused=True if this is the desired behavior.To further solve this, I set "allow_unused=True" in autograd. As a result, all my gradients go to 0. Why is this happening?

    opened by Sharut 5
Owner
Shivam Saboo
ML x Fashion @dresslife | Prev Yale, @intel, Interdigital
Shivam Saboo
GitHub
H2O is an Open Source, Distributed, Fast & Scalable Machine Learning Platform: Deep Learning, Gradient Boosting (GBM) & XGBoost, Random Forest, Generalized Linear Modeling (GLM with Elastic Net), K-Means, PCA, Generalized Additive Models (GAM), RuleFit, Support Vector Machine (SVM), Stacked Ensembles, Automatic Machine Learning (AutoML), etc.

H2O H2O is an in-memory platform for distributed, scalable machine learning. H2O uses familiar interfaces like R, Python, Scala, Java, JSON and the Fl

H2O.ai 6.1k Jan 5, 2023
The official implementation of You Only Compress Once: Towards Effective and Elastic BERT Compression via Exploit-Explore Stochastic Nature Gradient.

You Only Compress Once: Towards Effective and Elastic BERT Compression via Exploit-Explore Stochastic Nature Gradient (paper) @misc{zhang2021compress,

null 46 Dec 7, 2022
Official implementation of the ICML2021 paper "Elastic Graph Neural Networks"

ElasticGNN This repository includes the official implementation of ElasticGNN in the paper "Elastic Graph Neural Networks" [ICML 2021]. Xiaorui Liu, W

liuxiaorui 34 Dec 4, 2022
Code for "Learning Structural Edits via Incremental Tree Transformations" (ICLR'21)

Learning Structural Edits via Incremental Tree Transformations Code for "Learning Structural Edits via Incremental Tree Transformations" (ICLR'21) 1.

NeuLab 40 Dec 23, 2022
The official PyTorch code for 'DER: Dynamically Expandable Representation for Class Incremental Learning' accepted by CVPR2021

DER.ClassIL.Pytorch This repo is the official implementation of DER: Dynamically Expandable Representation for Class Incremental Learning (CVPR 2021)

rhyssiyan 108 Jan 1, 2023
Implementation of the paper "Self-Promoted Prototype Refinement for Few-Shot Class-Incremental Learning"

Self-Promoted Prototype Refinement for Few-Shot Class-Incremental Learning This is the implementation of the paper "Self-Promoted Prototype Refinement

Kai Zhu 78 Dec 2, 2022
Official repository of the paper 'Essentials for Class Incremental Learning'

Essentials for Class Incremental Learning Official repository of the paper 'Essentials for Class Incremental Learning' This Pytorch repository contain

null 33 Nov 27, 2022
[TPAMI 2021] iOD: Incremental Object Detection via Meta-Learning

Incremental Object Detection via Meta-Learning To appear in an upcoming issue of the IEEE Transactions on Pattern Analysis and Machine Intelligence (T

Joseph K J 66 Jan 4, 2023
The code repository for "PyCIL: A Python Toolbox for Class-Incremental Learning" in PyTorch.

PyCIL: A Python Toolbox for Class-Incremental Learning Introduction • Methods Reproduced • Reproduced Results • How To Use • License • Acknowledgement

Fu-Yun Wang 258 Dec 31, 2022
Official Implementation of CVPR 2022 paper: "Mimicking the Oracle: An Initial Phase Decorrelation Approach for Class Incremental Learning"

(CVPR 2022) Mimicking the Oracle: An Initial Phase Decorrelation Approach for Class Incremental Learning ArXiv This repo contains Official Implementat

Yujun Shi 24 Nov 1, 2022