0x 00：空格
()
\cite
Table. \ref
Fig. \ref
Alg. \ref
.
0x01：全文统一写法
MNIST、CIFAR10、ImageNet
Sign-OPT
变量符号及含义
…, …., and ….
0x02：算法中的专有名词
建议去查算法原版所述 wiki 可查
eg：line-search candidate
0x03：间距
全文紧凑
0x04：图、表
图的caption在下面
表的caption在上面
0x05：符号
矢量加粗，脚标不加粗
标量不加粗
0x06：缩写
i.e.
a.k.a.
w.r.t
0x07：参考文献
1、作者，标题（第一个单词的字符首字母大写），In Proceedings of 会议（每一个单词的首字母大写，除了介词），pages多少页到多少页，年份
eg:1
Yanpei Liu, Xinyun Chen, Chang Liu, and Dawn Song. Delving into transferable adversarial examples and black-box attacks. In Proceedings of the International Conference on Learning Representations, pages 1–14, 2017.
2、作者，标题（第一个单词首字母大写），期刊（杂志）（每一个单词的首字母大写，除了介词），vol，pages，year
3、同一会议 期刊的名字要全文统一。
0x08：摘要（Abstract）
however
we find 我们的部分差不多占一半
0x09：引言
1、一般三段到四段
第一段背景
第二段过去的技术方法 缺点
第三段 motivated by balalbaba/ Based on these finding 我们提出了吧啦吧
2、刚好占满第一页
0x10：实验
1、实验中各个参数的设置 要交代清楚
2、算法：一般参数放在最前面 然后是input output
3、#号 是the number of 的意思
0x11：a、the的用法。
有限定的前面加the
0x12：logical word
| due to |
| meanwhile |
| thereby |
| by contrast |
| thus |
| developed |
| respectively |
| Since |
| utterly |
| extra |
| causing |
| actually |
| carry out |
| Unfortunately |
| unavailable |
| a large number of |
| Therefore |
| As a result |
| Obviously |
| Hence |
| That is to say |
| even if |
0x13：nice word
| 搜索 |
|---|
| constrain |
| challenging |
| adversary |
| makes great advances |
| is employed to |
| obtain/find |
| costs |
| a huge amount of |
| resulting in |
| degradation |
| termed |
| enhance |
| requires |
| achieve |
| significantly |
| overall |
| achieving tremendous attention |
| focuses on |
| realistic scenario |
| dramatically |
| worked out |
| Inversely |
| superiority |
| demonstrate |
| illustration |
| below |
| as follows |
| visualizes |
| among which |
| be activated to |
| figure out |
| Note that |
| denotes |
| dramatically surpasses\ |
| verify |
| in terms of |
| confirmed |