Knowledge Consolidation based Class Incremental Online Learning with Limited Data

This problem setting which necessitates a class incremental learning approach; is more challenging than standard class incremental learning (2) Data for each class is given in an online [Javed and White, 2019] due to additional constraints: (1) fashion, i.e., each training example is seen only once Data in each class appears in the online fashion, i.e., the model during training; (3) Each class has very few training sees every training example exactly once; (2) The number of examples; and (4) We do not use or assume access training examples in each class is very small; and (3) We do to any replay/memory to store data from previous not use any replay/memory to store the training examples from classes. Therefore, in this setting, we have to handle previous classes. This is the most general setting for class incremental twofold problems of catastrophic forgetting and learning and various practical usage scenario can be overfitting. In our approach, we learn robust representations obtained through this or a relaxed setting. For instance, in face that are generalizable across tasks without recognition, it is common to have few examples per class but suffering from the problems of catastrophic forgetting usually not in an online learning fashion, whereas for a robot and overfitting to accommodate future classes navigating in an environment, the setting would also be online.