Efficient Multi-Model Fusion with Adversarial Complementary Representation Learning
This work addresses performance inefficiencies in multi-model systems for machine learning tasks, offering an incremental improvement over existing fusion techniques.
The paper tackled the problem of redundancy in multi-model fusion for tasks like speaker verification and image classification, proposing an adversarial complementary representation learning framework that improved performance more efficiently than traditional methods, with experimental results validating its efficacy.
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 distinct, complementary representations. We make three detailed explanations of why this works and experimental results demonstrate that our method more efficiently improves performance compared to traditional MMF. Furthermore, attribution analysis validates the model trained under ACoRL acquires more complementary knowledge, highlighting the efficacy of our approach in enhancing efficiency and robustness across tasks.