Temporal Imbalance of Positive and Negative Supervision in Class-Incremental Learning
This addresses the challenge of stable long-term learning for visual tasks with dynamically evolving data, though it is incremental as it builds on existing CIL methods by focusing on a specific overlooked factor.
The paper tackles the problem of catastrophic forgetting in Class-Incremental Learning by identifying temporal imbalance as a key cause of prediction bias, and proposes Temporal-Adjusted Loss (TAL) to mitigate this, resulting in significant reductions in forgetting and improved performance on benchmarks.
With the widespread adoption of deep learning in visual tasks, Class-Incremental Learning (CIL) has become an important paradigm for handling dynamically evolving data distributions. However, CIL faces the core challenge of catastrophic forgetting, often manifested as a prediction bias toward new classes. Existing methods mainly attribute this bias to intra-task class imbalance and focus on corrections at the classifier head. In this paper, we highlight an overlooked factor -- temporal imbalance -- as a key cause of this bias. Earlier classes receive stronger negative supervision toward the end of training, leading to asymmetric precision and recall. We establish a temporal supervision model, formally define temporal imbalance, and propose Temporal-Adjusted Loss (TAL), which uses a temporal decay kernel to construct a supervision strength vector and dynamically reweight the negative supervision in cross-entropy loss. Theoretical analysis shows that TAL degenerates to standard cross-entropy under balanced conditions and effectively mitigates prediction bias under imbalance. Extensive experiments demonstrate that TAL significantly reduces forgetting and improves performance on multiple CIL benchmarks, underscoring the importance of temporal modeling for stable long-term learning.