Amazon EC2 可以掛多個 Elastic Inference 了

看到 Jeff Barr 的 tweet:

所以是一台 Amazon EC2 的主機可以掛多個 Elastic Inference (GPU) 了,這主要應該還是對現有的使用者有幫助。還沒有使用的應該會往新的 AWS Inferentia 測試?(參考「AWS 開始推自己的 Machine Learning Chip」)

企業內的文件搜尋系統 Amazon Kendra

AWS 推出了具有語意分析的能力,可以直接丟自然語言進去搜尋的 Amazon Kendra:「Announcing Amazon Kendra: Reinventing Enterprise Search with Machine Learning」。

之前 Google 有推出過 Google Search Appliance 也是做企業內資料的整合 (2016 年收掉了),但應該沒有到可以用自然語言搜尋?

Amazon Kendra 的費用不算便宜,Enterprise Edition 提供 150GB 的容量與 50 萬筆文件,然後提供大約 40k query/day,這樣要 USD$7/hr,一個月大約是 USD$5,040,不過對於企業來說應該是很有用...

另外有提到這邊 query 收費的部份是估算,會依照 query 問題的難易度而不同:

Actual queries per day will vary based on query complexity, which greatly varies from customer to customer. Less complex queries (e.g. “leave policy”) consume less resources to run, and more complex queries (e.g. “What’s the daily parking allowance in Seattle?”) consume more resources to run. The total number of queries you can run with your allocated resources will depend on your mix of queries. The max queries per day provided above is an estimate, assuming 80% less complex queries and 20% more complex queries.

這樣頗有趣的,感覺可以處理簡單的分析了?

Amazon Detective:用 Machine Learning 分析可能的安全問題

也是這次 AWS re:Invent 發表的服務,透過 Machine Learning 分析可能的安全問題:「Introducing Amazon Detective」。

透過現有的各種 log 建立模型分析:

Amazon Detective can analyze trillions of events from multiple data sources such as Virtual Private Cloud (VPC) Flow Logs, AWS CloudTrail, and Amazon GuardDuty, and automatically creates a unified, interactive view of your resources, users, and the interactions between them over time.

依照 log 的量算錢的,然後 preview 階段不收費,所以有興趣的人可以開起來跑看看?

AWS 開始推自己的 Machine Learning Chip

除了常見的 GPU 類,以及之前公佈過的 FPGA 外,這次 AWS 推出的是自己做的晶片 AWS Inferentia,以及對應到 EC2 上的機種 inf1:「Amazon EC2 Update – Inf1 Instances with AWS Inferentia Chips for High Performance Cost-Effective Inferencing」。

從介紹可以看到支援的形式:

Each AWS Inferentia chip supports up to 128 TOPS (trillions of operations per second) of performance at low power to enable multiple chips per EC2 instance. AWS Inferentia supports FP16, BF16, and INT8 data types. Furthermore, Inferentia can take a 32-bit trained model and run it at the speed of a 16-bit model using BFloat16.

然後常見的框架都先弄好支援了:

AWS Inferentia comes with the AWS Neuron software development kit (SDK) that enables complex neural net models, created and trained in popular frameworks to be executed using AWS Inferentia based EC2 Inf1 instances. Neuron consists of a compiler, run-time, and profiling tools and is pre-integrated into popular machine learning frameworks including TensorFlow, Pytorch, and MXNet to deliver optimal performance of EC2 Inf1 instances.

現在看起來類似於 Google 弄的 TPU,專為 machine learning 搞出來的 ASIC,等一陣子應該就會有兩者的比較了...

AWS 展示了 DeepComposer

今年 AWSre:Invent 又開始了,照慣例有很多東西會在會場上發表 (尤其是現場表演起來會很炫的),其中一個是 AWS DeepComposer:「AWS DeepComposer – Compose Music with Generative Machine Learning Models」。

在現場有人錄影下來放到 Twitter 上可以直接看:

現場展示了輸入一段旋律,而 AWS DeepComposer 可以補上其他樂器的配樂。在 blog 上的介紹文章也可以看到同樣的說明:

  • Log into the DeepComposer console,
  • Record a short musical tune, or use a prerecorded one,
  • Select a generative model for your favorite genre, either pretrained or your own,
  • Use this model to generate a new polyphonic composition,
  • Play the composition in the console,
  • Export the composition or share it on SoundCloud.

就... 很炫 XD

Amazon Aurora 可以直接使用 AWS 的 Machine Learning 服務

AWS 宣佈了 Amazon Aurora 可以直接使用 AWS 自家的 Machine Learning 服務:「New for Amazon Aurora – Use Machine Learning Directly From Your Databases」。

整合了兩個服務,分別是 Amazon SageMaker (各類的模型) 以及 Amazon Comprehend (文字處理相關)。

目前只有 Amazon Aurora MySQL 5.7 的版本有支援,其他的還在做:

The new machine learning integration is available today for Aurora MySQL 5.7, with the SageMaker integration generally available and the Comprehend integration in preview. You can learn more in the documentation. We are working on other engines and versions: Aurora MySQL 5.6 and Aurora PostgreSQL 10 and 11 are coming soon.

這個整合讓程式用起來更方便了...

Amazon Redshift 會自動在背景重新排序資料以增加效能

Amazon Redshift 的新功能,會自動在背景重新排序資料以增加效能:「Amazon Redshift introduces Automatic Table Sort, an automated alternative to Vacuum Sort」。

版本要到更新到 1.0.11118,然後預設就會打開:

This feature is available in Redshift 1.0.11118 and later.

Automatic table sort is now enabled by default on Redshift tables where a sort key is specified.

重新排序的運算會在背景處理,另外帶一些行為學習分析:

Redshift runs the sorting in the background and re-organizes the data in tables to maintain sort order and provide optimal performance. This operation does not interrupt query processing and reduces the compute resources required by operating only on frequently accessed blocks of data. It prioritizes which blocks of table to sort by analyzing query patterns using machine learning.

算是丟著讓他跑就好的東西,升級上去後可以看一下 CloudWatch 的報告,這邊沒有特別講應該是還好... XD

擋 Live 與 Podcast 內廣告的工具

看到「An adblocker for live radio streams and podcasts. Machine learning meets Shazam.」這個專案,這個把 machine learning 用到「正途」上了啊...

不過畢竟是比較複雜的演算法,會吃不少 CPU 資源:

On a regular laptop CPU and with the Python time-frequency analyser, computations run at 5-10X for files and at 10-20% usage for live stream.

不過看用法還是偏向 library 性質,如果要大力推廣可能還是需要有其他人包個更好的界面...

用 Machine Learning 改善 Streaming 品質的服務與論文

Hacker News 上看到「Puffer」這個服務,裡面利用了 machine learning algorithm 動態調整 bitrate,以提昇傳輸品質。

測試得到的數據後來被整理起來一起放進論文:「Continual learning improves Internet video streaming」。

在開頭介紹了 Fugu 這個演算法:

We describe Fugu, a continual learning algorithm for bitrate selection in streaming video.

而 Puffer 就是實驗網站:

We evaluate Fugu with Puffer, a public website we built that streams live TV using Fugu and existing algorithms. Over a nine-day period in January 2019, Puffer streamed 8,131 hours of video to 3,719 unique users.

這個站台提供了許多真實的頻道進行測試:

Stream live TV in your browser. There's no charge. You can watch U.S. TV stations affiliated with the NBC, CBS, ABC, PBS, FOX, and Univision networks.

可以看到 Fugu 的結果很好,比起其他提出的方案是全面性的改善:

這邊是用 WebSocket 測試,並且配合了不同的 TCP congestion algorithm,沒有太考慮額外的計算成本...

AI 版的星海爭霸二將直接透過歐洲區的 Battle.net 匿名與人類對戰

前幾天 Blizzard 公佈的消息,DeepMind 的星海爭霸二 AI (AlphaStar) 將會透過 Blizzard 的 Battle.net 歐洲區伺服器跟人類對戰:「DeepMind Research on Ladder」。

Experimental versions of DeepMind’s StarCraft II agent, AlphaStar, will soon play a small number of games on the competitive ladder in Europe as part of ongoing research into AI.

預設是不會對到的,需要選擇參與:

If you would like the chance to help DeepMind with its research by matching against AlphaStar, you can opt in by clicking the “opt-in” button on the in-game popup window. You can alter your opt-in selection at any time by using the “DeepMind opt-in” button on the 1v1 Versus menu.

但你仍然不會知道對手是人還是 AI,而且如同一般對戰情況,這會影響到你的戰績:

For scientific test purposes, DeepMind will be benchmarking AlphaStar’s performance by playing anonymously during a series of blind trial matches. This means the StarCraft community will not know which matches AlphaStar is playing, to help ensure that all games are played under the same conditions. AlphaStar plays with built-in restrictions that the DeepMind team has defined in consultation with pro players. A win or a loss against AlphaStar will affect your MMR as normal.

okay,這樣大概知道為什麼只開放歐洲區了...

加州從今年七月開始,禁止 AI 偽裝成人類 (前幾天也有一些新聞在報導):「A California law now means chatbots have to disclose they’re not human」,對應的法條在「Bill Text - SB-1001 Bots: disclosure」這邊可以看到:

17941. (a) It shall be unlawful for any person to use a bot to communicate or interact with another person in California online, with the intent to mislead the other person about its artificial identity for the purpose of knowingly deceiving the person about the content of the communication in order to incentivize a purchase or sale of goods or services in a commercial transaction or to influence a vote in an election. A person using a bot shall not be liable under this section if the person discloses that it is a bot.

(b) The disclosure required by this section shall be clear, conspicuous, and reasonably designed to inform persons with whom the bot communicates or interacts that it is a bot.

而加州是 Blizzard Entertainment 的總部...

法條上面對「online platform」有設計排除條款,不過如果只算星海二的人數,有可能不到這個豁免限制... 所以得避開而改用歐洲區來測試?

(c) “Online platform” means any public-facing Internet Web site, Web application, or digital application, including a social network or publication, that has 10,000,000 or more unique monthly United States visitors or users for a majority of months during the preceding 12 months.

(c) This chapter does not impose a duty on service providers of online platforms, including, but not limited to, Web hosting and Internet service providers.

美國軍方應該是超級關注這個議題,相較於 AlphaGo 或是 AlphaZero 是資訊完全透明的遊戲,這次要踏入非對稱資訊的遊戲。

如果在這個領域上有成果的話,可以預期未來的戰爭 (yeah 實體戰爭) 會開始大量採用 AI 了...