Bilevel Online Deep Learning in Non-stationary Environment

TL;DR

This paper introduces Bilevel Online Deep Learning (BODL), a framework designed to adapt neural networks to non-stationary data streams with concept drift, using ensemble classifiers, bilevel optimization, and drift detection for improved performance.

Contribution

The paper proposes a novel BODL framework combining bilevel optimization and online ensemble classifiers to effectively handle concept drift in streaming data.

Findings

BODL outperforms existing methods on various datasets.

The ensemble approach improves robustness to concept drift.

The drift detection mechanism effectively triggers model updates.

Abstract

Recent years have witnessed enormous progress of online learning. However, a major challenge on the road to artificial agents is concept drift, that is, the data probability distribution would change where the data instance arrives sequentially in a stream fashion, which would lead to catastrophic forgetting and degrade the performance of the model. In this paper, we proposed a new Bilevel Online Deep Learning (BODL) framework, which combine bilevel optimization strategy and online ensemble classifier. In BODL algorithm, we use an ensemble classifier, which use the output of different hidden layers in deep neural network to build multiple base classifiers, the important weights of the base classifiers are updated according to exponential gradient descent method in an online manner. Besides, we apply the similar constraint to overcome the convergence problem of online ensemble framework. Then an effective concept drift detection mechanism utilizing the error rate of classifier is designed to monitor the change of the data probability distribution. When the concept drift is detected, our BODL algorithm can adaptively update the model parameters via bilevel optimization and then circumvent the large drift and encourage positive transfer. Finally, the extensive experiments and ablation studies are conducted on various datasets and the competitive numerical results illustrate that our BODL algorithm is a promising approach.