Continual learning has attracted increasing research attention in recent years due to its promising experimental results in real-world applications. In this paper, we study the issue of calibration in continual learning which reliably quantifies the uncertainty of model predictions. Conformal prediction (CP) provides a general framework for model calibration, which outputs prediction intervals or sets with a theoretical high coverage guarantee as long as the samples are exchangeable. However, the tasks in continual learning are learned in sequence, which violates the principle that data should be exchangeable. Meanwhile, the model learns the current task with limited or no access to data from previous tasks, which is not conducive to constructing the calibration set. To address these issues, we propose a CP-based method for model uncertainty quantification in continual learning (CPCL), which also reveals the connection between prediction interval length and forgetting. We analyze the oracle prediction interval in continual learning and theoretically prove the asymptotic coverage guarantee of CPCL. Finally, extensive experiments on simulated and real data empirically verify the validity of our proposed method.

Popular deep-learning models in the field of image classification suffer from catastrophic forgetting—models will forget previously acquired skills when learning new ones. Generative replay (GR), which typically consists of a generator and a classifier, is an efficient way to mitigate catastrophic forgetting. However, conventional GR methods only focus on a single instruction relationship (generator-to-classifier), where the generator synthesizes samples for previous tasks to instruct the training of the classifier, while ignoring the ways in which the classifier can benefit the generator. In addition, most generative replay methods typically reuse the generated samples to update the generator, which causes the samples regenerated by the generator deviating from the distribution of previous tasks. To overcome these two issues, we propose a novel approach, called deep diffusion-based generative replay (DDGR), which adopts a diffusion model as the generator and calculates an instruction-operator through the classifier to instruct the generation of samples. Extensive experiments in class incremental (CI) and class incremental with repetition (CIR) settings demonstrate the advantages of DDGR. Our code is available at https://github.com/xiaocangshengGR/DDGR.

Abstract This paper studies the couplet generation model which automatically generates the second line of a couplet by giving the first line. Unlike other sequence generation problems, couplet generation not only considers the sequential context within a sentence line but also emphasizes the relationships between the corresponding words of first and second lines. Therefore, a trapezoidal context character embedding the vector model has been developed firstly, which considers the ‘sequence context’ and the ‘corresponding word context’ simultaneously. Afterwards, we chose the typical encoder–decoder framework to solve the sequence–sequence problems, of which the encoder and decoder are used by bi-directional GRU and GRU, respectively. In order to further increase the semantic consistency of the first and second lines of couplets, the pre-trained sentence vector of the first line is added to the attention mechanism in the model. To verify the effectiveness of the method, it is applied to the real data set. Experimental results show that our proposed model can compete with the up-to-date methods, and both adding sentence vectors to attention and using trapezoidal context character vectors can improve the effectiveness of the algorithm.