Efficient Feature Transformations for Discriminative and Generative Continual Learning

Verma, Vinay Kumar; Liang, Kevin J; Mehta, Nikhil; Rai, Piyush; Carin, Lawrence

doi:10.1109/cvpr46437.2021.01365

Cited by 42 publications

(15 citation statements)

References 19 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Singh et al [152] further extended [151] to be used for both zero-shot and non-zero-shot continual learning. Verma et al [153] proposed an Efficient Feature Transformation (EFT) to separate the shared and the task-specific features using Efficient Convolution Operations [154].…”

Section: Dynamic Network Architecturesmentioning

confidence: 99%

Recent Advances of Continual Learning in Computer Vision: An Overview

Qu,

Rahmani,

et al. 2021

Preprint

View full text Add to dashboard Cite

show abstract

Section: Dynamic Network Architecturesmentioning

confidence: 99%

Recent Advances of Continual Learning in Computer Vision: An Overview

Qu,

Rahmani,

et al. 2021

Preprint

View full text Add to dashboard Cite

show abstract

“…They also proposed a less-forget constraint to retain the features of the old samples and the parameters of the classifier by limiting the angle of the features of the sample on the old and new models. Verma et al [19] applied a Kullback-Leibler divergence loss function to expand the decision boundaries between the features in the data of the given task. In this work, the inter-class information is used to constrain the update of model parameters to ensure that the model can grasp the multi-granularity correlations of classes to learn old and new classes effectively.…”

Section: Class Incremental Learning Methodsmentioning

confidence: 99%

Multi-Granularity Regularized Re-Balancing for Class Incremental Learning

Chen

Wang

2022

IEEE Trans. Knowl. Data Eng.

View full text Add to dashboard Cite

Deep learning models suffer from catastrophic forgetting when learning new tasks incrementally. Incremental learning has been proposed to retain the knowledge of old classes while learning to identify new classes. A typical approach is to use a few exemplars to avoid forgetting old knowledge. In such a scenario, data imbalance between old and new classes is a key issue that leads to performance degradation of the model. Several strategies have been designed to rectify the bias towards the new classes due to data imbalance. However, they heavily rely on the assumptions of the bias relation between old and new classes. Therefore, they are not suitable for complex real-world applications. In this study, we propose an assumption-agnostic method, Multi-Granularity Regularized re-Balancing (MGRB), to address this problem. Re-balancing methods are used to alleviate the influence of data imbalance; however, we empirically discover that they would under-fit new classes. To this end, we further design a novel multi-granularity regularization term that enables the model to consider the correlations of classes in addition to re-balancing the data. A class hierarchy is first constructed by grouping the semantically or visually similar classes. The multi-granularity regularization then transforms the one-hot label vector into a continuous label distribution, which reflects the relations between the target class and other classes based on the constructed class hierarchy. Thus, the model can learn the inter-class relational information, which helps enhance the learning of both old and new classes. Experimental results on both public datasets and a real-world fault diagnosis dataset verify the effectiveness of the proposed method.

show abstract

“…Xiang et al [17] proposed a learning strategy based on conditional adversarial networks using memory-efficient statistical information to store old knowledge and a discriminator to distinguish between real and generated examples. EFTs [18] proposes a task-specific feature mapping transformation strategy that significantly reduces the computational cost of generating models. GANMemory [19] proposes sequential style modulation on a good base GAN model to form a targeted sequential generative model to achieve incremental learning.…”

Section: Related Workmentioning

confidence: 99%

“…Current incremental learning can be broadly classified into three categories depending on whether the training phase requires real old classes data [9], [10], [11], [12], [13] or generated old classes data [14], [15], [16], [17], [18], [19], or does not use any old classes data [20], [21], [22], [23], [24], [25], [26].…”

mentioning

confidence: 99%

Adaptive Threshold Hierarchical Incremental Learning Method

et al. 2023

IEEE Access

View full text Add to dashboard Cite

Traditional deep convolutional neural networks have achieved excellent performance on various machine learning tasks. Still, they perform poorly in continuous data stream environments, where models trained on new datasets often suffer from a significant drop in performance on old datasets, a phenomenon known as ''catastrophic forgetting.'' Incremental learning can help solve the ''catastrophic forgetting'' problem in deep learning by learning new knowledge while retaining what has already been learned. In practice, incremental learning algorithms usually need to be deployed on edge devices with limited memory and restricted access to training data, facing the problems of high model complexity and imbalance between old and new categories of data. We propose an Adaptive Threshold Hierarchical Incremental Learning (ATHIL) method to address the above problems. Our proposed method does not require additional data and model storage space during the training process, combines local weight discrete coefficient thresholding and the mean nearest neighbor principle, uses a sparse matrix hierarchical masking network, and flexibly adjusts the network structure according to different tasks to achieve learning multiple image classification tasks in a single network. The experimental results show that the performance of the proposed method significantly outperforms existing methods on fine-grained classification datasets under three evaluation metrics.

show abstract

Efficient Feature Transformations for Discriminative and Generative Continual Learning

Cited by 42 publications

References 19 publications

Recent Advances of Continual Learning in Computer Vision: An Overview

Recent Advances of Continual Learning in Computer Vision: An Overview

Multi-Granularity Regularized Re-Balancing for Class Incremental Learning

Adaptive Threshold Hierarchical Incremental Learning Method

Contact Info

Product

Resources

About