… feature map transformation for largescale continual learning, which we call Efficient Feature
Transformation (… Finally, we propose a strategy for maximizing feature distance to improve …

… , online continual learning is … learning on a stream of tasks, which is ubiquitous in realistic
scenarios. Thus, in this work, we focus on the continual learning setting in an online learning …

… (1) This paper proposes to employ PTMs in the continual learning scenario with different
tuning strategies. However, further exploration of the performance of different PTMs, such as …

… DualNet for the online continual learning settings [36… In the batch continual learning setting
[46], the model is allowed to … Particularly, instead of generating the transformation coefficients …

… We show that Tukeys transformation significantly reduces the skewness of the distributions
… transformation, we observe that the new class features are comparatively better clustered. …

… We offer an alternative approach here where private features are induced by the feature
transformation strategy of [21] and the adversarial game is played by one and only one …

… could be reprogrammed with a task-dependent linear transformation. The main contributions
of this work are: • We propose a novel continual learning solution for CNNs, which involves …

… transformation follows each feature extractor and then all the … of continual learning due to
the addition of all old feature … learned feature extractor and only the pruned feature extractors …

… Therefore, we propose to learn a linear transformation that projects the new feature
space back to the old one and in the following section, we show it can achieve both goals. …

… study continual learning for reconstructionstyle tasks such as these. We also examine the
performance of continual … is used as applying an affine transformation on the output of the point …