Continual learning paper repository

This repository contains an incomplete (but dynamically updated) list of papers exploring continual learning in machine learning and neuroscience, and accompanies the related paper Towards continual task learning in artificial neural networks: current approaches and insights from neuroscience.

Machine learning reviews, surveys, & tutorials
Classic papers
Architectural approaches to continual learning
Regularisation
Training regime
Neuroscience
Citation

arXiv link

The full paper complementing this repository is available at https://arxiv.org/abs/2112.14146

Machine learning

Reviews, surveys, & tutorials

Title	Link	Relevance
Continual lifelong learning with neural networks: A review	Neural Networks	•••
Catastrophic forgetting in connectionist networks	TICS	•••
Neuroscience-Inspired Artificial Intelligence	Neuron	••
How to grow a mind: Statistics, structure, and abstraction	Science	•
Deep learning	Nature	•
Universal Intelligence: A Definition of Machine Intelligence	arXiv	•

Classic papers

Title	Link	Relevance
Catastrophic forgetting in connectionist networks	TICS	•••
Catastrophic Interference in Connectionist Networks: The Sequential Learning Problem	Psychology of Learning & Motivation	•••
Connectionist models of recognition memory: Constraints imposed by learning and forgetting functions	Psychological Review	•

Architectural approaches to continual learning

Title	Link	Implementation
Progressive Neural Networks	arXiv	PyTorch TensorFlow
Neurogenesis deep learning: Extending deep networks to accommodate new classes	IEEE	–
Adaptive structural learning of artificial neural networks	ICML	TensorFlow

Regularisation

Title	Link	Implementation
Learning without forgetting	arXiv	PyTorch
Distilling the knowledge in a neural network	arXiv	PyTorch TensorFlow
Overcoming catastrophic forgetting in neural networks	arXiv	PyTorch TensorFlow
Note on the quadratic penalties in elastic weight consolidation	PNAS	–
Measuring catastrophic forgetting in neural networks	arXiv	–
Continual learning through synaptic intelligence	ICML	TensorFlow
An Empirical Investigation of Catastrophic Forgetting in Gradient-Based Neural Networks	arXiv	Theano

Training regime

Title	Link	Implementation
How transferable are features in deep neural networks?	arXiv	Caffe
CHILD: A First Step Towards Continual Learning	Machine Learning	–
Curriculum learning	ICML	–
Continual Learning with Deep Generative Replay	arXiv	PyTorch
Experience Replay for Continual Learning	arXiv	–
Brain-inspired replay for continual learning with artificial neural networks	Nature Communications	PyTorch
REMIND Your Neural Network to Prevent Catastrophic Forgetting	arXiv	PyTorch

Neuroscience

Title	Link	Relevance
Regulation and function of adult neurogenesis: from genes to cognition	Physiological Review	•
When and where do we apply what we learn?: A taxonomy for far transfer	Psychological Bulletin	•
Does the hippocampus map out the future?	TICS	•
Organizing conceptual knowledge in humans with a gridlike code	Science	•
Song replay during sleep and computational rules for sensorimotor vocal learning	Science	•
A theory of the discovery and predication of relational concepts	Psychological Review	•
Preplay of future place cell sequences by hippocampal cellular assemblies	Nature	••
Comparing continual task learning in minds and machines	PNAS	••
Cascade models of synaptically stored memories	Neuron	•
Selective suppression of hippocampal ripples impairs spatial memory	Nature Neuroscience	••
The analogical mind	MIT Press	•
What learning systems do intelligent agents need? Complementary learning systems theory updated	TICS	•
Human replay spontaneously reorganizes experience	Cell	•••
Compartmentalized dendritic plasticity and input feature storage in neurons	Nature	•
Neuroconstructivism: How the brain constructs cognition	Oxford University Press	•
Why there are complementary learning systems in the hippocampus and neocortex: insights from the successes and failures of connectionist models of learning and memory	Psychological Review	•••
The Role of Hippocampal Replay in Memory and Planning	Current Biology	••
Brain imaging of language plasticity in adopted adults: Can a second language replace the first?	Cerebral Cortex	•••
Memory formation: Let’s replay	Elife	•
Strengthening individual memories by reactivating them during sleep	Science	••
Replay of neuronal firing sequences in rat hippocampus during sleep following spatial experience	Science	•
Crossmodal spatial attention	Annals of the NY Academy of Sciences	•
The merging of the senses	MIT Press	•
Development of multisensory integration from the perspective of the individual neuron	Nature Reviews Neuroscience	••
The hippocampal indexing theory and episodic memory: updating the index	Hippocampus	••
Stably maintained dendritic spines are associated with lifelong memories	Nature	•••

Citation

BibTeX

@misc{mccaffary2021continual,
      title={Towards continual task learning in artificial neural networks: current approaches and insights from neuroscience}, 
      author={David McCaffary},
      year={2021},
      eprint={2112.14146},
      archivePrefix={arXiv},
      primaryClass={cs.LG}
}

Repository of continual learning papers

Related tags

Overview

Continual learning paper repository

arXiv link

Machine learning

Reviews, surveys, & tutorials

Classic papers

Architectural approaches to continual learning

Regularisation

Training regime

Neuroscience

Citation

BibTeX

Owner

A Semantic Segmentation Network for Urban-Scale Building Footprint Extraction Using RGB Satellite Imagery

Convert Pytorch model to onnx or tflite, and the converted model can be visualized by Netron

This repo contains the code and data used in the paper "Wizard of Search Engine: Access to Information Through Conversations with Search Engines"

Deeplab-resnet-101 in Pytorch with Jaccard loss

A disassembler for the RP2040 Programmable I/O State-machine!

SpeechBrain is an open-source and all-in-one speech toolkit based on PyTorch.

A privacy-focused, intelligent security camera system.

Monocular Depth Estimation - Weighted-average prediction from multiple pre-trained depth estimation models

Github for the conference paper GLOD-Gaussian Likelihood OOD detector

Paddle implementation for "Highly Efficient Knowledge Graph Embedding Learning with Closed-Form Orthogonal Procrustes Analysis" (NAACL 2021)

Code for our paper at ECCV 2020: Post-Training Piecewise Linear Quantization for Deep Neural Networks

PyTorch Implementation of Unsupervised Depth Completion with Calibrated Backprojection Layers (ORAL, ICCV 2021)

Adaout is a practical and flexible regularization method with high generalization and interpretability

[ECE NTUA] 👁 Computer Vision - Lab Projects & Theoretical Problem Sets (2020-2021)

GLIP: Grounded Language-Image Pre-training

Approaches to modeling terrain and maps in python

Train emoji embeddings based on emoji descriptions.

Easy Parallel Library (EPL) is a general and efficient deep learning framework for distributed model training.

It is a simple library to speed up CLIP inference up to 3x (K80 GPU)

Implementation of Cross Transformer for spatially-aware few-shot transfer, in Pytorch