PruneFuse: Efficient Data Selection via Weight Pruning and Network Fusion
This addresses the need for efficient data selection to enhance training efficiency and minimize annotation requirements in deep learning, though it appears incremental as it builds on pruning and fusion techniques.
The paper tackles the problem of high computational costs in data selection for deep neural networks by introducing PruneFuse, which uses pruned networks for selection and fuses them with the original network, resulting in reduced computational costs and better performance than baselines.
Efficient data selection is crucial for enhancing the training efficiency of deep neural networks and minimizing annotation requirements. Traditional methods often face high computational costs, limiting their scalability and practical use. We introduce PruneFuse, a novel strategy that leverages pruned networks for data selection and later fuses them with the original network to optimize training. PruneFuse operates in two stages: First, it applies structured pruning to create a smaller pruned network that, due to its structural coherence with the original network, is well-suited for the data selection task. This small network is then trained and selects the most informative samples from the dataset. Second, the trained pruned network is seamlessly fused with the original network. This integration leverages the insights gained during the training of the pruned network to facilitate the learning process of the fused network while leaving room for the network to discover more robust solutions. Extensive experimentation on various datasets demonstrates that PruneFuse significantly reduces computational costs for data selection, achieves better performance than baselines, and accelerates the overall training process.