Efficient Multi-Model Fusion with Adversarial Complementary Representation Learning

arXiv:2404.15704v1 Announce Type: new
Abstract: Single-model systems often suffer from deficiencies in tasks such as speaker verification (SV) and image classification, relying heavily on partial prior knowledge during decision-making, resulting in suboptimal performance. Although multi-model fusion (MMF) can mitigate some of these issues, redundancy in learned representations may limits improvements. To this end, we propose an adversarial complementary representation learning (ACoRL) framework that enables newly trained models to avoid previously acquired knowledge, allowing each individual component model to learn maximally …

abstract adversarial arxiv classification cs.ai cs.lg cs.sd decision eess.as fusion image improvements knowledge making performance prior redundancy representation representation learning speaker systems tasks type verification