CLeaR: An Adaptive Continual Learning Framework for Regression Tasks

Abstract

Catastrophic forgetting means that a trained neural network model gradually forgets the previously learned tasks when retrained on new tasks. Overcoming the forgetting problem is a major problem in machine learning. Numerous continual learning algorithms are very successful in incremental classification tasks, where new labels appear frequently. However, there is currently no research that addresses the catastrophic forgetting problem in regression tasks as far as we know. This problem has emerged as one of the primary constraints in some applications, such as renewable energy forecasts. This article clarifies the problem-related definitions and proposes a new methodological framework that can forecast regression task targets and update itself by continual learning. The framework consists of forecasting neural networks and buffers, which store newly collected data from a data stream in an application. Changes in the probability distribution of the data stream will be identified by the framework and learned sequentially. The framework is called CLeaR (Continual Learning for Regression Tasks), where components can be flexibly customized for a specific application scenario. We design two sets of experiments to evaluate the CLeaR framework concerning fitting error (training), prediction error (test), and forgetting ratio. The first one is based on an artificial time series to explore how hyperparameters affect the CLeaR framework. The second one is designed with data collected from European wind farms to evaluate the performance of the CLeaR framework in a real-world application. The experimental results demonstrate that the CLeaR framework can efficiently accumulate knowledge in the data stream and improve the prediction accuracy. The article concludes with further research issues arising from requirements to extend the framework.

Adaptive Explainable Continual Learning Framework for Regression Problems with Focus on Power Forecasts

“…Farquhar et al introduce five core desiderata for evaluating CL algorithms and designing classification experiments [5]. In previous work [13], we give five suggestions for designing CL regression experiments:…”

“…For answering this question, my research will focus on novelty detection (concept shift and drift) using deterministic and probabilistic methods. For example, in [13], the trigger condition depends on the number of newly collected novel samples. Besides, updating could also be triggered due to the estimation of new samples' entropy.…”

“…However, more specific metrics are required for evaluating the updated models comprehensively in the CL setting. In [12,13], the models are evaluated in terms of fitting error, prediction error, and forgetting ratio. We also consider training time as a significant evaluation factor [11], especially in real-time applications.…”

“…Besides, due to concept drift in the data stream, the parameters of the used scaler might have to be updated, for example, the maximum and minimum of the min-max scaler or the mean and variance of the standard scaler. The CLeaR framework is a continual learning framework based on buffered data [13] that is grouped as novelty and familiarity by the deterministic or probabilistic novelty detector and stored in the corresponding buffer. The novelty from the infinite data stream indicates what the trained model cannot predict accurately and should continually learn.…”

“…To analyze the shortcomings of common experimental setups as well as restrictions of general evaluation metrics; 5. To summarize relevant research challenges being faced for our proposed CL framework [13] and develop it further; 6. To develop visualization utilities and propose comprehensive evaluation metrics to make CL explainable; 7.…”