ReinDSplit: Reinforced Dynamic Split Learning for Pest Recognition in Precision Agriculture

2025-06-16

AI Summary

Agricultural edge devices for insect monitoring exhibit significant heterogeneity in computation, energy, and connectivity; conventional static DNN partitioning leads to straggler bottlenecks, resource underutilization, and accuracy degradation. Method: We propose a Q-learning–based dynamic DNN partitioning framework that models partition-point selection as a finite-state Markov decision process (MDP), enabling adaptive, heterogeneous-device-aware splitting for ResNet18, GoogLeNet, and MobileNetV2. Contribution/Results: The framework jointly optimizes privacy preservation, low latency, and high accuracy. On three insect datasets, MobileNetV2 achieves 94.31% classification accuracy. It significantly improves edge resource utilization, system scalability, and long-term participatory sustainability—overcoming the limitations of one-size-fits-all partitioning paradigms.

Technology Category

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

Application Category

• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications
• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs
• Economics, Online Markets and Human Computation: Incentives in network design for Web infrastructures and ecosystems

Abstract

To empower precision agriculture through distributed machine learning (DML), split learning (SL) has emerged as a promising paradigm, partitioning deep neural networks (DNNs) between edge devices and servers to reduce computational burdens and preserve data privacy. However, conventional SL frameworks' one-split-fits-all strategy is a critical limitation in agricultural ecosystems where edge insect monitoring devices exhibit vast heterogeneity in computational power, energy constraints, and connectivity. This leads to straggler bottlenecks, inefficient resource utilization, and compromised model performance. Bridging this gap, we introduce ReinDSplit, a novel reinforcement learning (RL)-driven framework that dynamically tailors DNN split points for each device, optimizing efficiency without sacrificing accuracy. Specifically, a Q-learning agent acts as an adaptive orchestrator, balancing workloads and latency thresholds across devices to mitigate computational starvation or overload. By framing split layer selection as a finite-state Markov decision process, ReinDSplit convergence ensures that highly constrained devices contribute meaningfully to model training over time. Evaluated on three insect classification datasets using ResNet18, GoogleNet, and MobileNetV2, ReinDSplit achieves 94.31% accuracy with MobileNetV2. Beyond agriculture, ReinDSplit pioneers a paradigm shift in SL by harmonizing RL for resource efficiency, privacy, and scalability in heterogeneous environments.

Problem

• Dynamic DNN splitting for heterogeneous edge devices
• Optimizing split learning efficiency without accuracy loss
• Balancing workloads across devices using reinforcement learning

Innovation

• Reinforcement learning-driven dynamic DNN split optimization
• Q-learning agent adaptively balances device workloads
• Markov decision process ensures efficient split layer selection