日常记录
新的一周新的开始,中间blog断更了几天,是我状态不太好,也没有能够积极调整。总之现在已经完全恢复了,继续加油努力!
早上起来晨跑10km,上午背单词、练口语,中午在学习micropython,下午晚上看论文找资料。
学习记录
- 看论文 Fast 2D filter with low false positive for network packet inspection
看了这篇文章加上网上的一些资料,才算是对商过滤器真正的理解,之前在看布隆过滤器的时候并没有看懂。文章里给的图,非常直观,对理解有很大帮助。
canonical entry: Depicts the entry in which a fingerprint’s remainder would be saved in the lack of collisions. run: Comprises whole the entries that save remainders that have the identical quotient, whereas cluster: includes the longest series of taken entries in which only the first element has been saved in its canonical entry. A cluster contains at least one run. The combination of the cluster’s first element with the associated 3 bits in every entry gives the ability to recover the fingerprint of every saved remainder inside the cluster. These associated 3 bits are: is_occupied: if the entry is the canonical entry for any element inside the filter this bit position is set to 1; otherwise, it is set to 0. is_continuation: Set to 1 if the entry contains a remainder that is not the first of the run. is_shifted: Set to 1 to depict that the saved remainder is not in its canonical entry. When there is an element to be mapped into QF, the quotient is used as an index to the entry and the is_occupied bit is set to 1. Whenever there is soft collision, any remainder will be shifted, if necessary, and the other two bits are set appropriately [39].
对这篇文章最大的感受就是他的语言阅读起来非常的困难,很多长难句、各种各样的从句以及有一些我从来都没有听过的说法,下面的问题记录会详细记录。
问题记录
- 在introduction中这个".ida"是什么?不过这个应该不是重点
- 在双 BF 字符串匹配模块中,没有理解这个"slight sufficient"为什么能够使FP为0,emmm这个我再想想
- 第一次见"oscillate this defect"这种说法。
- 这个poison函数是没看懂咋得到的
问题解决
反思
今天还是比较充实的,虽然感觉不如最开始的状态,但也比前几天好太多了,逐步恢复。