论文标题
Protuner：使用Monte Carlo Tree搜索调整程序
ProTuner: Tuning Programs with Monte Carlo Tree Search
论文作者
论文摘要
我们探索在众所周知的艰巨任务中应用蒙特卡洛树搜索（MCT）算法：用于高性能深度学习和图像处理的计划。我们在卤化物之上构建框架，并表明MCT可以胜过最先进的Beam-Search算法。与Beam搜索不同，该搜索是由贪婪的中级性能比较在部分和有意义的时间表之间进行的，MCT会比较完整的时间表，并在做出任何中级计划决策之前进行前进。我们进一步探讨了对标准MCTS算法的修改，并将实际执行时间测量与成本模型相结合。我们的结果表明，MCT可以在16个真实基准的套件上胜过梁搜索。
We explore applying the Monte Carlo Tree Search (MCTS) algorithm in a notoriously difficult task: tuning programs for high-performance deep learning and image processing. We build our framework on top of Halide and show that MCTS can outperform the state-of-the-art beam-search algorithm. Unlike beam search, which is guided by greedy intermediate performance comparisons between partial and less meaningful schedules, MCTS compares complete schedules and looks ahead before making any intermediate scheduling decision. We further explore modifications to the standard MCTS algorithm as well as combining real execution time measurements with the cost model. Our results show that MCTS can outperform beam search on a suite of 16 real benchmarks.
