Continual Hyperbolic Learning of Instances and Classes

Continual learning has traditionally focused on classifying either instances or classes, but real-world applications, such as robotics and self-driving cars, require models to handle both simultaneously. To mirror real-life scenarios, we introduce the task of continual learning of instances and classes, at the same time. This task challenges models to adapt to multiple levels of granularity over time, which requires balancing fine-grained instance recognition with coarse-grained class generalization. In this paper, we identify that classes and instances naturally form a hierarchical structure. To model these hierarchical relationships, we propose HyperCLIC, a continual learning algorithm that leverages hyperbolic space, which is uniquely suited for hierarchical data due to its ability to represent tree-like structures with low distortion and compact embeddings. Our framework incorporates hyperbolic classification and distillation objectives, enabling the continual embedding of hierarchical relations. To evaluate performance across multiple granularities, we introduce continual hierarchical metrics. We validate our approach on EgoObjects, the only dataset that captures the complexity of hierarchical object recognition in dynamic real-world environments. Empirical results show that HyperCLIC operates effectively at multiple granularities with improved hierarchical generalization. Continual learning addresses a long-standing challenge in machine learning: learning from new classes often leads to catastrophic forgetting of old classes (Kirkpatrick et al., 2017; Wu et al., 2019; Magistri et al., 2024; Lyle et al., 2024; De Lange et al., 2021; Wang et al., 2024). To mitigate this, numerous solutions have been proposed, including data replay (Bang et al., 2021; Wang et al., 2021), regularization (Yin et al., 2021; Lee et al., 2020), and knowledge distillation (Kang et al., 2022; Dong et al., 2022). While these methods primarily focus on class-level discrimination, fewer works have expanded the scope to instance-level continual learning. In robotics, for example, classifying specific instances of objects enables robots to make informed decisions about their use or placement (Ammirato, 2019; Singh et al., 2014; Held et al., 2016).