Classroom-Inspired Multi-Mentor Distillation with Adaptive Learning Strategies

TL;DR

ClassroomKD introduces a dynamic multi-mentor knowledge distillation framework inspired by classroom teaching, which adaptively selects mentors and strategies based on their effectiveness to improve model performance across tasks.

Contribution

The paper presents a novel framework with dynamic mentor selection and adaptive distillation strategies, outperforming existing methods in image classification and pose estimation tasks.

Findings

Outperforms existing knowledge distillation methods on CIFAR-100 and ImageNet.

Effective in 2D human pose estimation on COCO and MPII datasets.

Adaptive mentor selection improves knowledge transfer efficiency.

Abstract

We propose ClassroomKD, a novel multi-mentor knowledge distillation framework inspired by classroom environments to enhance knowledge transfer between the student and multiple mentors with different knowledge levels. Unlike traditional methods that rely on fixed mentor-student relationships, our framework dynamically selects and adapts the teaching strategies of diverse mentors based on their effectiveness for each data sample. ClassroomKD comprises two main modules: the Knowledge Filtering (KF) module and the Mentoring module. The KF Module dynamically ranks mentors based on their performance for each input, activating only high-quality mentors to minimize error accumulation and prevent information loss. The Mentoring Module adjusts the distillation strategy by tuning each mentor's influence according to the dynamic performance gap between the student and mentors, effectively modulating the learning pace. Extensive experiments on image classification (CIFAR-100 and ImageNet) and 2D human pose estimation (COCO Keypoints and MPII Human Pose) demonstrate that ClassroomKD outperforms existing knowledge distillation methods for different network architectures. Our results highlight that a dynamic and adaptive approach to mentor selection and guidance leads to more effective knowledge transfer, paving the way for enhanced model performance through distillation.