Optimizing Deep Learning for Skin Cancer Classification: A Computationally Efficient CNN with Minimal Accuracy Trade-Off

2025-05-27

AI Summary

To address the high computational overhead hindering the deployment of skin cancer classification models on mobile and edge devices, this paper proposes a custom lightweight CNN architecture optimized for both accuracy and efficiency on the HAM10000 dataset. The model achieves near-parity performance with ResNet50—suffering only a marginal 0.022% drop in classification accuracy—while reducing parameter count by 96.7% (to 0.692M) and FLOPs by 99.25% (to 30.04M). This design substantially lowers memory footprint, energy consumption, and inference latency, effectively bridging the longstanding trade-off between deep learning model complexity and clinical deployability. The proposed solution thus enables practical, real-time skin cancer screening in resource-constrained environments.

Technology Category

• Machine Learning: Learning on the Edge & Model Compression
• Computer Vision: Learning & Optimization for CV
• Search and Optimization: Learning to Search

Application Category

• User Modeling, Personalization and Recommendation: On-Device user modeling, personalization, and recommendation
• Responsible Web: Human-perceived consequences of algorithmic deployment on the web
• Web Mining and Content Analysis: Large pretrained models with web data

Abstract

The rapid advancement of deep learning in medical image analysis has greatly enhanced the accuracy of skin cancer classification. However, current state-of-the-art models, especially those based on transfer learning like ResNet50, come with significant computational overhead, rendering them impractical for deployment in resource-constrained environments. This study proposes a custom CNN model that achieves a 96.7% reduction in parameters (from 23.9 million in ResNet50 to 692,000) while maintaining a classification accuracy deviation of less than 0.022%. Our empirical analysis of the HAM10000 dataset reveals that although transfer learning models provide a marginal accuracy improvement of approximately 0.022%, they result in a staggering 13,216.76% increase in FLOPs, considerably raising computational costs and inference latency. In contrast, our lightweight CNN architecture, which encompasses only 30.04 million FLOPs compared to ResNet50's 4.00 billion, significantly reduces energy consumption, memory footprint, and inference time. These findings underscore the trade-off between the complexity of deep models and their real-world feasibility, positioning our optimized CNN as a practical solution for mobile and edge-based skin cancer diagnostics.

Problem

• Reducing computational overhead in skin cancer classification models
• Balancing accuracy and efficiency for resource-constrained environments
• Proposing a lightweight CNN alternative to transfer learning models

Innovation

• Custom CNN reduces parameters by 96.7%
• Lightweight CNN minimizes FLOPs and energy
• Optimized CNN for mobile cancer diagnostics