Towards Oracle Knowledge Distillation with Neural Architecture Search

By Minsoo Kang, Jonghwan Mun, Bohyung Han. 2019

49 citations recorded by the source corpus.

We present a novel framework of knowledge distillation that is capable of learning powerful and efficient student models from ensemble teacher networks. Our approach addresses the inherent model capacity issue between teacher and student and aims to maximize benefit from teacher models during distillation by reducing their capacity gap. Specifically, we employ a neural architecture search technique to augment useful structures and operations, where the searched network is appropriate for knowledge distillation towards student models and free from sacrificing its performance by fixing the network capacity. We also introduce an oracle knowledge distillation loss to facilitate model search and distillation using an ensemble-based teacher model, where a student network is learned to imitate oracle performance of the teacher. We perform extensive experiments on the image classification datasets—CIFAR-100 and TinyImageNet—using various networks. We also show that searching for a new student model is effective in both accuracy and memory size and that the searched models often outperform their teacher models thanks to neural architecture search with oracle knowledge distillation.

Subfields: Artificial Intelligence, Computer Vision and Pattern Recognition

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

1. Dacheng Tao
2. Jianping Gou
3. S. Maybank
4. Baosheng Yu
5. Chenglong Bao
6. Linfeng Zhang
7. Kaigui Bian
8. Yushuo Guan

In Computer Vision and Pattern Recognition

1. Dacheng Tao
2. Jianping Gou
3. S. Maybank
4. Cong Yao
5. Kaigui Bian
6. Yushuo Guan
7. Baosheng Yu
8. Chenglong Bao

Citations in the corpus

• Knowledge Distillation: A Survey (2020, methodology): “To address this issue, neural architecture search has been adopted to find the appropriate student architecture in knowledge distillation, such as oracle-based (Kang et al., 2019) and architecture-aware knowledge distillation (Liu et al.”
• Self-Distillation: Towards Efficient and Compact Neural Networks (2021, methodology): “Recently, researchers find that the choice of teacher models has a great impact on the accuracy of student models and the teacher with the highest accuracy is not the best teacher for distillation [32], [33], [34].”
• Differentiable Feature Aggregation Search for Knowledge Distillation (2020, background): “KDAS [15] builds up the student network progressively based on an ensemble of independently learned student networks.”