Green AI approach optimizes Android malware detection with INT8 quantization

Researchers have developed a Green AI approach to optimize Android malware detection, addressing the trade-off between security and energy consumption on mobile devices. By comparing standard FP32 models with INT8 quantized neural networks, they found that INT8 quantization significantly reduces model size and energy usage while maintaining high detection accuracy. Specifically, shallow quantized architectures like 3-layer and 4-layer QNNs proved effective in lowering energy costs by improving throughput and reducing CPU high-power state duration, enabling efficient malware protection on resource-constrained smartphones. AI

IMPACT Enables more energy-efficient and effective AI-driven security solutions on mobile devices.

RANK_REASON Academic paper detailing a new methodology for AI model optimization. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.AI →

• android
• Drebin
• INT8
• multilayer perceptron
• QNNs
• TUANDROMD

• paper
• other

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →