
放棄巴點睇SNN![]()
堆proof too long didnt read
放棄巴點睇SNN![]()
堆proof too long didnt read
新野嚟?有冇link?
突發媽呀 個supervisor覆左cold email
佢下星期香港有conference話可以見我
![]()
![]()
佢咩U?你極想跟佢?係就開始狂J佢papers
當時第一封cold email就咁講research interest
第二封我就睇左佢大約一半同重點既paper 寫左兩個research proposal send比佢
跟住就有reply了![]()
媽呀 我淨係cold email左佢一個 因為淨係想跟佢 我諗住冇reply就下年再算![]()
不過佢係你地睇唔起澳洲 melb u
放棄巴點睇SNN![]()
堆proof too long didnt read
新野嚟?有冇link?
Self-Normalizing Neural Networks
遲啲睇睇
太長唔quote太多![]()
以下講嘅我唔太肯定啱唔啱
通常討論sequence of random variables(in particular markov chain),我哋會將佢諗成一串sequence of numbers
而個dynamic就係將串數字移左或者移右一格
例如將
(x1,x2,x3,....)
依串數字shift一shift變
(x2,x3,x4,...)
ergodicity就話你知你不斷咁樣shift法,你會見到哂所有嘅"random pattern"(因為咁shift法會行勻成個space), 個randomness就measured by你個probability
你可以想像如果個markov chain要滿足咁嘅條件,佢就應該唔會係periodic,transient之類
[quote][quote][quote]
太長唔quote太多![]()
以下講嘅我唔太肯定啱唔啱
通常討論sequence of random variables(in particular markov chain),我哋會將佢諗成一串sequence of numbers
而個dynamic就係將串數字移左或者移右一格
例如將
(x1,x2,x3,....)
依串數字shift一shift變
(x2,x3,x4,...)
ergodicity就話你知你不斷咁樣shift法,你會見到哂所有嘅"random pattern"(因為咁shift法會行勻成個space), 個randomness就measured by你個probability
你可以想像如果個markov chain要滿足咁嘅條件,佢就應該唔會係periodic,transient之類
做咩派膠比你諗起果個人