Energy-based Latent Aligner for Incremental Learning
This addresses the problem of forgetting in incremental learning for AI systems, offering a plug-and-play module that is incremental but provides complementary advantages to existing methods.
The paper tackles catastrophic forgetting in deep learning models during incremental learning by proposing ELI, an energy-based latent aligner that learns a manifold to align latent representations, resulting in consistent improvements across datasets and over 5% accuracy gain when added to a state-of-the-art object detector.
Deep learning models tend to forget their earlier knowledge while incrementally learning new tasks. This behavior emerges because the parameter updates optimized for the new tasks may not align well with the updates suitable for older tasks. The resulting latent representation mismatch causes forgetting. In this work, we propose ELI: Energy-based Latent Aligner for Incremental Learning, which first learns an energy manifold for the latent representations such that previous task latents will have low energy and the current task latents have high energy values. This learned manifold is used to counter the representational shift that happens during incremental learning. The implicit regularization that is offered by our proposed methodology can be used as a plug-and-play module in existing incremental learning methodologies. We validate this through extensive evaluation on CIFAR-100, ImageNet subset, ImageNet 1k and Pascal VOC datasets. We observe consistent improvement when ELI is added to three prominent methodologies in class-incremental learning, across multiple incremental settings. Further, when added to the state-of-the-art incremental object detector, ELI provides over 5% improvement in detection accuracy, corroborating its effectiveness and complementary advantage to existing art.