QSplitFL: Capability Aware Deep Q-Learning for Optimal Split Point Selection in Split Federated Learning

In three linesQSplitFL introduces a Deep Q-Network framework for automatic optimal split point selection in Split Federated Learning. Using lightweight hardware metrics (CPU, memory, battery, network latency) instead of model weights, the committee-based DQN architecture improves convergence on MNIST, Fashion-MNIST, CIFAR-10/100 with CNN, ResNet50, MobileNetV4, ConvNeXt.

Summary generated by Claude — human-verified