Symbolic Integration by Integrating Learning Models with Different Strengths and Weaknesses

Integration is indispensable, not only in mathematics, but also a wide range of other fields. A deep learning method has recently been developed and shown to be capable integrating mathematical functions that could previously integrated on computer. However, treats integration as equivalent natural language translation does reflect information. In this study, we adjusted the model take information into account models learn order numerical operations more robustly. way, achieved 98.80% correct answer rate with symbolic integration, higher than any existing method. We judged correctness based whether derivative primitive function was consistent integrand. By building an strategy, 99.79% answers integration.