希望下一次meeting時(shí)自己的口語(yǔ)能有大幅度地提高
下次開(kāi)會(huì)記得錄音利职，積累口語(yǔ)表達(dá)趣效，以及下來(lái)復(fù)盤瘦癌！
本次暑研的主要任務(wù)：
看一看TensorFlow的應(yīng)用是否能跑在oneOS上猪贪？
-
在本地跑一個(gè)TensorFlow的程序，
- 不必start from scratch,可以拿開(kāi)源項(xiàng)目的代碼
- realistic讯私； raw data是從sensor中得到的热押，實(shí)時(shí)數(shù)據(jù) e.g. 利用溫度濕度來(lái)預(yù)測(cè)明天的天氣；利用公路上攝像頭得到的斤寇；uber用gps的data 去預(yù)測(cè)traffic pattern桶癣，比如說(shuō)什么時(shí)候某個(gè)地方會(huì)有很多人；twitter有個(gè)程序去判斷是人還是機(jī)器娘锁。主要是放在論文中牙寞，選TensorFlow是因?yàn)槭莻€(gè)大家都知道的平臺(tái)，同時(shí)要選擇具有實(shí)際應(yīng)用價(jià)值的例子。
=>因此 1st week plan 找合適的項(xiàng)目间雀，和kumseok交流悔详。emm建議用word或者latex寫(xiě)好發(fā)過(guò)去
-
break it down into smaller pieces, say, several parts, data, filter the raw data and change their format, feed it into a train model, finally, get an output. Change the local program into a network version. maybe transfer them directly to another PC is the simplist way, but no one uses that. Can you find some ways that are cost-effective? there are transfer cost, bandwidth etc.
-
Try your program on some platforms, say, the IoT platform of Amazon,aws.?? maybe this is the most suitable one. Or others like the IoT platform of Microsoft, Samsung.
-
Try this on OneOS
OK, the first problem for me is to get familiar with the tensorflow and run the program on my PC.
Here are some general ideas of OneOS.
As we know, nowadays, if we want to write a program for IoT devices of certain brand, we have to use the corresponding platform. say, for IoT devices of Amazon, we have to use aws...for coding. So, now we want to build a platform that allows to write the program using a united platform. That is One OS.
Also, we want to gather the IoT devices and make them form a cluster, using distributed system, edge computing.
From now on, I plan to practice my English everyday and I hope I can finish this task by Monday. Come on.
