Targeted at PPT:
- whether to write the name of Professor Huang on the first page of PPT
- From engineering perspective to introduce the problem?
- Introduction of Info-Clustering and Metric Page has a little more interval pages
- research works seem not belonging to yourself.
- You should draw a figure to illustrate your main contribution.
Splitted Time: 3min Background 5min Related work 17min my work 3min summary and future plan
建议用演示模式,可以演示动画
- P3: 图说不清楚
- P4: 可以加上动画,说明聚类过程
- P5: 引用加上作者,不然不知所指(不够专业,如何指望别人能找到你说的工作)
- P6: 可能不能在这么短时间内了解你的算法,考虑加动画,构造更简单的例子
- P7: 你为啥要用B_{ij}这个记号。tilde都可以省略,能简则简。你写的公式不能指望别人先行忽略。
- P8: 单看这一页感觉只是简单罗列,毕竟听众完全不了解这些方法
- P9: 注意用简单的方式表述主要的信息,听众不会记得公式细节
- P10: 你这样说wij = 1/2 ||bij||^2 不如说 wij = I(Xi; Xj)。后者直观表示了互信息。
- P12: 有没有可能举例子说明性质?以及,这些性质为什么重要?
- P15: 到这里我已经不记得gamma是啥了,
- P16: 公式排版格式不一。很难让读者读完你slides上的文字。
- P21: 这一页很可能会快速跳过
问题背景解决三件事,为什么做聚类,为什么做系统聚类,为什么用信息论做系统聚类。
background 3min
related work 5min
my work 17min
summary and future work 3min
One big problem: Infomation Centric Part is limited to part of theoretical part, in application the weight is computed from some similarity measures.
Another improvement: collect COVID 19 data and do some clustering work.