Multi-view Contrastive Learning for Online Knowledge Distillation
This work addresses the limitation of existing OKD methods in neglecting representational knowledge, offering a domain-specific improvement for image classification tasks.
The paper tackles the problem of online knowledge distillation (OKD) by proposing Multi-view Contrastive Learning (MCL) to capture feature correlations from multiple peer networks, resulting in improved classification performance with large margins over state-of-the-art methods without extra inference cost.
Previous Online Knowledge Distillation (OKD) often carries out mutually exchanging probability distributions, but neglects the useful representational knowledge. We therefore propose Multi-view Contrastive Learning (MCL) for OKD to implicitly capture correlations of feature embeddings encoded by multiple peer networks, which provide various views for understanding the input data instances. Benefiting from MCL, we can learn a more discriminative representation space for classification than previous OKD methods. Experimental results on image classification demonstrate that our MCL-OKD outperforms other state-of-the-art OKD methods by large margins without sacrificing additional inference cost. Codes are available at https://github.com/winycg/MCL-OKD.