CCAPs:Convolutional Capsule networks(technical details)
time：pm15:00-15:20 March 13,2021
place:广东蕉岭丘成桐国际会议中心（广东省数学会2021年会）
Reporter:Aixiang(Andy) Chen
Abstract: Capsule network overcomes the fuzzy location of features brought about by pooling operation used in traditional convolutional neural network, making it have the ability of object recognition based on spatial position reasoning . On the basis of the reasoning and recognition principle of capsule network, this talk clearly shows the technical details of the capsule network through a simple example(manual calculation step by step), including the dynamic routing process, its convergence effect as well.
CCAPs:卷积胶囊网络（技术细节）
报告时间：2021年3月19日 上午10:00
地点：北2-501（人工智能与深度学习研究所讨论班）
报告人：陈蔼祥
摘要：胶囊网络克服了传统卷积网络池化操作带来的特征所在位置模糊的不足，使之具备了空间位置推理识别的能力。本报告介绍胶囊网络推理识别原理基础上，通过一个简单的手工算例逐步清晰地展示胶囊网络的各个技术细节，包含其核心的动态路由过程及其迭代的收敛效果。
