Abstract:The rapid evolution of deep learning has led to significant achievements in computer vision, primarily driven by complex convolutional neural networks (CNNs). However, the increasing depth and parameter count of these networks often result in overfitting and elevated computational demands. Knowledge distillation (KD) has emerged as a promising technique to address these issues by transferring knowledge from a large, well-trained teacher model to a more compact student model. This paper introduces a novel knowledge distillation method that simplifies the distillation process and narrows the performance gap between teacher and student models without relying on intricate knowledge representations. Our approach leverages a unique teacher network architecture designed to enhance the efficiency and effectiveness of knowledge transfer. Additionally, we introduce a streamlined teacher network architecture that transfers knowledge effectively through a simplified distillation process, enabling the student model to achieve high accuracy with reduced computational demands. Comprehensive experiments conducted on the CIFAR-10 dataset demonstrate that our proposed model achieves superior performance compared to traditional KD methods and established architectures such as ResNet and VGG networks. The proposed method not only maintains high accuracy but also significantly reduces training and validation losses. Key findings highlight the optimal hyperparameter settings (temperature T = 15.0 and smoothing factor α = 0.7), which yield the highest validation accuracy and lowest loss values. This research contributes to the theoretical and practical advancements in knowledge distillation, providing a robust framework for future applications and research in neural network compression and optimization. The simplicity and efficiency of our approach pave the way for more accessible and scalable solutions in deep learning model deployment.

Knowledge Distillation with Deep Supervision

Deeply-Supervised Knowledge Distillation

Semi-Online Knowledge Distillation

Knowledge Distillation Meets Self-Supervision

Preparing Lessons: Improve Knowledge Distillation with Better Supervision

An Embarrassingly Simple Approach for Knowledge Distillation

Stage-by-stage Knowledge Distillation

Knowledge Distillation with a Precise Teacher and Prediction with Abstention

Deep Collective Knowledge Distillation

Knowledge Distillation Using Hierarchical Self-Supervision Augmented Distribution

Lightweight Self-Knowledge Distillation with Multi-source Information Fusion

SAKD: Sparse attention knowledge distillation

Better Teacher Better Student: Dynamic Prior Knowledge for Knowledge Distillation

Multi-target Knowledge Distillation Via Student Self-reflection

Obtain Dark Knowledge via Extended Knowledge Distillation.

Soft Hybrid Knowledge Distillation Against Deep Neural Networks

ResKD: Residual-Guided Knowledge Distillation

Highlight Every Step: Knowledge Distillation via Collaborative Teaching

TC<SUP>3</SUP>KD: Knowledge distillation via teacher-student cooperative curriculum customization

Simplified Knowledge Distillation for Deep Neural Networks Bridging the Performance Gap with a Novel Teacher–Student Architecture

Densely Guided Knowledge Distillation using Multiple Teacher Assistants