| Algorithm Acquisition Method based on Neural Networks with Monotone Versatile Aggregation Function |
 Dongho Lee
（指導教員：Tomonari Sei／
数理情報第4研究室）
Neural algorithmic reasoning is an approach to reason about natural data, which has a high-dimensional and rich representation, using neural networks. In this study, we investigate the learnability of algorithms by neural networks, and the improvement of learnability under the algorithm alignment framework. We explore on the field of algorithms and deep learning both to build powerful neural reasoner, especially focusing on Graph Neural Networks (GNNs) which align well with dynamic programming. We propose the generalization of some aggregate functions of GNNs based on the monotonicity assumption.
In this study, I was able to learn how to design the architecture of an artificial neural network to solve specific problem. Also, it was great experience to tackle about important parts and challenges in the design of neural networks for certain purposes.
|
