Unified Probabilistic Deep Continual Learning through Generative Replay and Open Set Recognition

Abstract

Modern deep neural networks are well known to be brittle in the face of unknown data instances and recognition of the latter remains a challenge. Although it is inevitable for continual-learning systems to encounter such unseen concepts, the corresponding literature appears to nonetheless focus primarily on alleviating catastrophic interference with learned representations. In this work, we introduce a probabilistic approach that connects these perspectives based on variational inference in a single deep autoencoder model. Specifically, we propose to bound the approximate posterior by fitting regions of high density on the basis of correctly classified data points. These bounds are shown to serve a dual purpose: unseen unknown out-of-distribution data can be distinguished from already trained known tasks towards robust application. Simultaneously, to retain already acquired knowledge, a generative replay process can be narrowed to strictly in-distribution samples, in order to significantly alleviate catastrophic interference.

arXiv commentSpecial Issue on Continual Learning in Computer Vision: Theory and Applications

References in corpus (33)

• Batch Normalization: Accelerating Deep Network Training by Reducing Internal Covariate Shift
• Auto-Encoding Variational Bayes
• Distilling the Knowledge in a Neural Network
• Overcoming catastrophic forgetting in neural networks
• Fashion-MNIST: a Novel Image Dataset for Benchmarking Machine Learning Algorithms
• Dropout as a Bayesian Approximation: Representing Model Uncertainty in Deep Learning
• A Baseline for Detecting Misclassified and Out-of-Distribution Examples in Neural Networks
• Pixel Recurrent Neural Networks
• Semi-Supervised Learning with Deep Generative Models
• Riemannian Walk for Incremental Learning: Understanding Forgetting and Intransigence
• Continual Learning Through Synaptic Intelligence
• A Simple Unified Framework for Detecting Out-of-Distribution Samples and Adversarial Attacks
• Can You Trust Your Model's Uncertainty? Evaluating Predictive Uncertainty Under Dataset Shift
• Autoencoding beyond pixels using a learned similarity metric
• Enhancing The Reliability of Out-of-distribution Image Detection in Neural Networks
• Towards Open World Recognition
• Adversarial Examples Are Not Bugs, They Are Features
• FearNet: Brain-Inspired Model for Incremental Learning
• Benchmarking Neural Network Robustness to Common Corruptions and Perturbations
• Variational Continual Learning
• Mnemonics Training: Multi-Class Incremental Learning without Forgetting
• Towards Robust Evaluations of Continual Learning
• IntroVAE: Introspective Variational Autoencoders for Photographic Image Synthesis
• Training Confidence-calibrated Classifiers for Detecting Out-of-Distribution Samples
• Dark Experience for General Continual Learning: a Strong, Simple Baseline
• Reducing Network Agnostophobia
• PixelVAE: A Latent Variable Model for Natural Images
• Life-Long Disentangled Representation Learning with Cross-Domain Latent Homologies
• Memory Replay GANs: learning to generate images from new categories without forgetting
• Do Deep Generative Models Know What They Don't Know?
• Disentangling Disentanglement in Variational Autoencoders
• A Unifying Bayesian View of Continual Learning
• Variational Autoencoder with Implicit Optimal Priors

Cited by in corpus (6)

• Task-agnostic Continual Learning with Hybrid Probabilistic Models
• Continual Learning via Bit-Level Information Preserving
• GAN Memory with No Forgetting
• CLEVA-Compass: A Continual Learning EValuation Assessment Compass to Promote Research Transparency and Comparability
• Resilience to the Flowing Unknown: an Open Set Recognition Framework for Data Streams
• A Procedural World Generation Framework for Systematic Evaluation of Continual Learning