分類
人工智慧

AI幫助全世界對抗COVID-19

今年的GTC是Nvidia迄今為止最大的盛事,但與世界其他地區一樣,它必須適應我們都發現的異常情況。Huang將他通常的大舞台換成九個片段,並用他的廚房作為背景。

人工智能正在協助全球進行COVID-19研究,其中大部分由NVIDIA GPU提供支持。這是一項艱鉅的任務,新藥的研發費用通常超過25億美元,每9年翻一番,而90%的努力卻失敗了。天擇軟件供應商

Nvidia希望在降低成本的同時幫助加快重要藥物的發現

Huang說:“(新工具)COVID-19迫在眉睫。”

Huang宣布推出NVIDIA Clara Discovery,這是一套可幫助科學家發現能夠挽救生命的新藥的工具。

NVIDIA Clara結合了影像學,放射學和基因組學,可幫助開發醫療保健AI應用程序。預先訓練的AI模型和特定於應用程序的框架可幫助研究人員找到目標,建立化合物並開發響應。

GSK首席科學官兼研發總裁Hal Barron博士評論說:

“人工智能和機器學習就像一台新的顯微鏡,將幫助科學家們看到他們原本無法看到的東西。

NVIDIA在計算方面的投資與深度學習的力量相結合,將為解決生命科學行業中一些最大挑戰的解決方案提供幫助,並幫助我們繼續為患者提供轉化醫學和疫苗。

與GSK在倫敦的新AI實驗室一起,我很高興現在可以使用這些先進技術來幫助英國傑出的科學家。”

由於自然語言處理技術的突破,研究人員現在可以使用特定於生物醫學的語言模型進行工作。這意味著研究人員可以組織和激活大型數據集,研究文獻,並對有關現有治療方法和其他重要現實世界數據的論文或專利進行分類。

Huang說:“在有流行的行業工具的地方,我們的計算機科學家會對其進行加速。” “在沒有工具的地方,我們會開發它們,例如NVIDIA Parabricks,Clara Imaging,BioMegatron,BioBERT,NVIDIA RAPIDS。”

我們都希望使用科學家可以使用的如此強大的新工具進行的COVID-19研究能夠在一兩年之內生產出疫苗,而通常情況下,它們需要花費十年甚至更長的時間才能製成。

“大數據,超級計算和人工智能的使用具有改變研發的潛力;從目標識別到臨床研究一直到新藥的推出,”阿斯利康(AstraZeneca)數據科學和AI負責人James Weatherall博士表示。

在他的主題演講中,黃仁勳提供了更多有關NVIDIA打造英國最快的超級計算機(將用於進一步的醫療研究-Cambridge-1)的努力的詳細信息。

NVIDIA已與領導對抗COVID-19和其他病毒的公司建立了合作夥伴關係,其中包括阿斯利康,GSK,倫敦國王學院,蓋伊和聖托馬斯NHS基金會信託基金以及初創公司Oxford Nanopore。這些合作夥伴可以利用Cambridge-1進行重要的研究。

Huang說:“要應對全球最緊迫的醫療保健挑戰,需要強大的計算資源來利用AI的功能。” “ Cambridge-1超級計算機將成為英國的創新中心,並進一步推動英國研究人員在關鍵醫療保健和藥物發現方面所做的開創性工作。”

而且,對於希望建立自己的AI超級計算機的組織,NVIDIA已宣布DGX SuperPOD作為世界上第一套交鑰匙AI基礎設施。該解決方案是針對NVIDIA在醫療保健,汽車,醫療保健,對話式AI,推薦系統,數據科學和計算機圖形學方面的研究成果而開發的。

雖然Huang的廚房不錯,但我敢肯定,他想回到GTC 2021主題演講的大舞台上。我們當然都希望COVID-19能夠在後視鏡中保持良好狀態。

※本文章屬於TNZE天擇集團所有嚴禁轉載※

分類
人工智慧

IBM最新AI更精準預測阿茲海默症

IBM開發了一種新的AI模型,該模型可以預測比標準臨床試驗更好的阿爾茨海默氏症的發作。

該AI被設計為非侵入性的,並使用來自給患者的口頭認知測試的短語言樣本。使用此樣本,AI模型能夠以71%左右的準確度預測老年癡呆症的發作。

為了進行比較,標準的臨床測試在大約59%的時間內正確無誤,診斷時間更長。當前的測試分析了人們隨著年齡增長而潛在的警告信號的描述能力。

該公司在一篇詳細介紹IBM模型的論文中說,它使用了Framingham心臟研究的數據。

這項研究始於1948年,涵蓋了建立AI來預測健康個體中無其他危險因素的阿爾茨海默氏病所需的多代人。對來自馬薩諸塞州及其家人的5,000名參與者進行了研究。

收集並分析了來自270位研究參與者的703個樣本,以創建一個數據集,該數據集由來自80個參與者的單個樣本組成,其中一半在達到85歲之前就出現了阿爾茨海默氏症。

在此數據集上對AI進行了訓練,以發現阿爾茨海默氏症的信號,例如單詞重複和使用語法結構較差的短句子。IBM的AI能夠在十個案例中的每七個案例中正確預測阿爾茨海默氏症的發作。

IBM打算使用更多數據來擴展其模型的培訓,以更好地反映社會,包括社會經濟,種族和地理因素。阿爾茨海默氏症的研究是IBM做出更大努力的一部分,旨在通過語音和語言中的生物標誌物和信號更好地理解神經系統健康和慢性疾病。

據估計,僅在美國,就有約550萬人患有阿爾茨海默氏病,一些研究表明,這是繼心髒病和癌症之後的第三大死亡原因。

儘管尚無治愈或預防阿爾茨海默氏病的方法,但早期診斷有助於盡可能多地為個人及其家人做準備。如果有治療方法,早發現的話幾乎可以肯定會更有效地治療老年癡呆症。

IBM在《柳葉刀》的科學雜誌EClinicalMedicine中發表了研究成果。輝瑞被披露為提供資金以從Framingham心臟研究聯合會獲取數據並支持IBM Research的參與。天擇線上系統

※本文章屬於TNZE天擇集團所有嚴禁轉載※

分類
人工智慧

人工智能可幫助患者,同時降低醫療人力需求

費恩斯坦研究所的一個團隊認為,人工智能可能是幫助患者獲得更多休息並減輕醫護人員負擔的關鍵。

每個人都知道充足的睡眠對恢復至關重要。但是,處於痛苦中的患者(或者像我這樣的失眠症患者)可能難以獲得所需的睡眠。

費恩斯坦研究所生物電子醫學研究所助理教授西奧多羅斯·扎諾斯(Theodoros Zanos)說:“休息是護理病人的關鍵要素,並且有據可查的是,睡眠中斷是一個常見的抱怨,可能會延遲出院和康復。”

當患者最終閉上眼睛時,他們想要的最後一件事就是醒來以檢查其生命體徵,但這種測量非常重要。

在《自然夥伴期刊》上發表的一篇論文中,研究人員詳細介紹了他們如何開發深度學習預測工具,該工具可以在一夜之間預測患者的穩定性。這樣可以防止進行多次不必要的檢查。

在2012年至2019年之間,對紐約諾斯韋爾健康醫院的213萬例患者就診的生命體徵指標用於訓練AI。數據包括心率,收縮壓,體溫,呼吸頻率和年齡。總共使用了2,430萬個生命體徵。

經過測試,人工智能在誤診一夜中誤診了10,000名患者中的兩名。研究人員指出,護士在常規診治中如何能夠解釋這兩個誤診病例。

根據該論文,大約有20%至35%的時間用於記錄患者的生命體徵。他們大約有10%的時間花在收集生命上。平均而言,護士目前必須每四到五小時收集一次病人的生命。

考慮到這一點,醫務人員感到如此繁重和壓力不足為奇。這些人希望提供他們可以提供的最佳護理,但只有兩隻手。使用AI騰出更多時間來履行其英雄職責,同時改善患者護理水平只是一件好事。

※本文章屬於TNZE天擇集團所有嚴禁轉載※

分類
人工智慧

Google解雇了道德操守的AI研究人員

批評公司之後,Google已解雇了道德AI開發領域的領軍人物。

Timnit Gebru被認為是該領域的先驅,並研究了大型語言模型中發現的風險和不平等現象。

Gebru聲稱她因未發表的論文而被Google解僱,並發送批評該公司行為的電子郵件。

論文質疑語言模型是否可能太大,誰能從中受益,以及它們是否會增加偏見和不平等現象。一般而言,最近的一些案例證實了她對大型模型和數據集的主張。

例如,麻省理工學院被迫於今年早些時候刪除了一個名為“ 8000萬個小圖像”的大型數據集。該數據集在訓練AI方面很受歡迎,但發現其中包含帶有種族主義,性別歧視和其他不可接受術語的圖像。

麻省理工學院網站上的一份聲明聲稱它沒有意識到冒犯性標籤,並且“這是依賴於WordNet名詞的自動數據收集程序的結果”。

聲明繼續解釋了數據集中包含的8000萬張圖像(尺寸僅為32×32像素),這意味著幾乎不可能進行手動檢查,並且不能保證所有令人反感的圖像都將被刪除。

據報導,Gebru向Google Brain女人和盟友列表服務器發送了一封電子郵件,該郵件“與Google經理的期望不一致”。

在電子郵件中,Gebru對Google在招聘女性方面缺乏進展感到沮喪。Gebru聲稱還被告知不要發表研究報告,並建議員工不要填寫多樣性文書,因為這沒關係。

除了被解僱的可疑原因外,格布魯還說,她的前同事收到了電子郵件,說她提出了辭職-但她聲稱情況並非如此:

Platformer從Google Research負責人Jeff Dean獲得了一封電子郵件,該電子郵件已發送給員工,並提供他對Gebru主張的支持:

“我們已經批准了Timnit和/或其他Google員工撰寫並發表的數十篇論文,但是如您所知,這些論文在內部審核過程中經常需要更改(甚至被認為不適合提交)。不幸的是,只有在截止日期前一天的通知中共享此特定論文-我們需要兩週的時間來進行此類審閱-然後,它沒有等待審閱者的反饋,而是被批准提交並提交。

然後,跨職能團隊審查了該文件,將其作為我們常規流程的一部分,並告知作者該文件不符合我們的出版標準,並得到了有關其原因的反饋。它忽略了太多相關的研究-例如,它談到了大型模型對環境的影響,但無視隨後的研究顯示出更高的效率。同樣,它引起了對語言模型偏見的擔憂,但沒有考慮到最近為減輕這些問題所做的研究。”

Dean繼續聲稱Gebru提出了要求,其中包括透露他和Google工程副總裁Megan Kacholia所諮詢的個人的身份,作為本文審閱的一部分。據報導,如果不能滿足要求,Gebru說她將離開公司。

這是一個反一個詞的情況,但是-對於一家已經在公眾和監管機構中引起質疑的公司而言,因其提出問題而被解僱的道德研究人員將不會成為良好的公關。

※本文章屬於TNZE天擇集團所有嚴禁轉載※

分類
人工智慧

基金公司以人工智能為手段,大膽向體育博彩市場叩關!

知名量化公司SIG (Susquehanna International Group)近期積極布局體育博彩市場,成立的新部門將
以發展人工智能技術針對籃球、橄欖球、足球等項目進行分析。事實上,借助人工智能分析體育博彩已非
新鮮事。據英國金融時報報導,Stratagen Technologies公司已在今年推出體育博彩基金,利用人工智
能進行投注。
知名量化公司SIG (Susquehanna International Group)最近成立了新部門,利用人工智慧對包括籃球、橄欖球、足
球和網球等體育博彩項目進行投注。
新部門名叫Nellie Analytics,位於都柏林,主要娶焦地緣政治分析和體育分析。SIG官網顯示,該部門正在招聘對體育
博彩市場有所瞭解的Python軟體工程師。
SIG戰略規劃主管 David Pollard在接受美國主流媒體採訪時稱: SIG的做市商背景有助於進行體交易。儘管目前美國
運動專案在歐洲博彩交易所的
交易量並不大,但體育博彩是一個很好的領域,SIG將關注交易量是否會增長。
不走尋常路的量化基金
SIG從來都是一家不走尋常路的量化基金公司。
博弈論的盛行在SIG由來已久。SIG認為,對於人們做的每件事來說,博弈論都是真實的。交易高手對於競爭、策略和
風控的追求,與博弈高手如出一辙。
正因如此,SIG向來重視遊戲。他們
不但用撲克作為培訓工具,甚至用撲克比赛作為校園招聘的手段,來測試大學生的
數學水準和風險控制能力。
除了捷克以外,棒球、象棋、魔術也是SIG十分偏爱的運動和遊戲項目。
體育博彩基金的興起
事實上,SIG並非唯一一家對體育博彩有興趣的基金公司。
據英國金融時報,前高盛交易員Charles McGarraugh創立的Stratagem Technologies公司今年推出了體育博彩基金。
這家公司致力於借助人工智慧來分析足球、籃球和網球比賽,並利用演算法在比賽前和比賽期間進行下注。
McGarraugh認為,體育博彩市場是檢驗真公司AI策略的最佳場合:
體育博彩市場規模很大,但十分分散,而體育比賽則是高度結構化和重複化的。
除此之外,2016年3月,澳大利亞一家名為Priomha的投資公司成立了一隻體夸博彩基金,更早的時候,2010年,位於
倫敦的Centaur Corporate公司成立了押注足球、養馬和網球比賽的體育博彩基金,兩年後因虧損而關閉

分類
人工智慧

AI時代下人工智慧與線上市場的完美結合

「人工智慧」的概念廣泛應用於現代生活中各種領域。AI時代的來臨,全面改變了各產業的運作模式與方
案。近幾年開始有不少國家紛紛將AI應用於線上博奕,不僅大幅提昇遊戲技能、創造大量商機,也能有效
地協助玩家克服賭癮,博彩娛樂更能細水長流,為博奕市場上的業者及玩家帶來雙贏的進展!
Artificial Intelligence in Online Gambling
What is Al’s role in gambling?
The concept of “artificial intelligence” is extensively used in various spheres of modern life, especially those
related to computer and Internet technologies. What it is and how it can be applied to the gambling industry-
all these questions are answered in Slotegrator’s current review.
The academic notion of “artificial intelligence” (AI) stands for the process of creation and development of
intellectual computer software. A distinctive feature of such software is the ability to process and analyze
information, as well as to accomplish intellectual and creative tasks.
First AI was applied to gambling in the middle of the last century. American cybernetic scientist Arthur Samuel
was the first one to create a type of software that allowed mainframe computers to play checkers with humans.
During the gameplay, the system was learning by itself, improving its gaming skills based on previous
experience.
In 1962, the program managed to beat R. Neely, the best USA checker player of that time. Particularly this
incident triggered further observations and AI software evolution.
Today, there exist a lot of ideas regarding Al development. Broadly speaking, they can be divided into two
types: the first one is based on a semiotic approach, and the second – on the biological one.
As for the first type, the program attempts to copy the human mind, while the second one makes use of natural
evolutionary algorithms, functioning as biological neural nets, resulting in intellectual activities.
Currently, a remarkable progress has been reached in development and integration of programs of the second
type. Consequently, a technology of machine learning, also known as Deep Learning, is successfully used in
iGaming.
The AI based programs are used in overcoming gambling addictions, retaining regular players, as well as for
scientific and educational purposes.
Inventions on current market
of LTTE cating
Today there are several effective solutions on the market. Last year BtoBet successfully presented its innovative
AI platform for online casinos. In a very short time, the software was tested on the global gambling market and
proved to be quite effectual.
The main task of the new software platform is to track player’ s actions and react to them, identifying potential
needs of the user. It is one of the most important tools of customer retention technique.
The platform responds to the player’s behavior in various environments and systems: social networks, mobile
applications, etc. By doing so, it is getting easier to meet the needs of the most capricious online casino visitors
quickly and efficiently, subsequently increasing online gambling popularity.
The AI concept was also used in poker for cognitive and scientific purposes. This game has always attracted
attention of various researchers and scientists, a fact that contributes to the online gaming popularity in
general.
Poker is known as a game based on incomplete data, since it implies randomness, while the number of possible
gaming combinations is indefinite. Texas Hold’em is one of the most common types of poker with practically
no limitations to the number of combinations.
The first full-fledged program for poker was launched and tested in 2015. Several world leading poker players
battled with the AI supercomputer Claudico.
The next famous developments of the AI were introduced by the scientists from the University of Alberta,
University of Charles and the Czech Technical University. During an experiment a system called DeepStack AI,
played around 44 852 times against 33 world’s strongest professional players, actual members of the
International Federation of Poker. Consequently, throughout all the games, the AI system showed successful
response results of 492 mbb/g (milli-big-blinds per game). The final figures are thought to be quite high-
about 100 mbb/g among experienced players.
Current Libratus systems created at the University of Carnegie Mellon and based on AI are considered as the
most effective ones. This system was internationally recognized as the one with the maximum effectiveness
after its landslide victory in games against humans during a 20-day poker tournament. The game of Texas
Hold’em was carried out in one of Pennsylvania’ s casinos within participation of world’s four best poker
players.
In recent years, many countries have been seriously considering the gambling addiction problem. In order to
prevent ludomania, lawmakers have been continuously introducing different bans on gambling activities,
measures that turn out to be not very beneficial for the development of the said industry. As it has been shown
in recent studies, the AI systems can be very effective in combating gambling addiction and its consequences.
Consequently, recent studies carried out by the University of London have a significant value for the entire
gaming industry. According to the scientists, they managed to elaborate a unique system based on the AI
allowing to detect a pathological addiction to gambling, even before it transforms into a real addiction.
In order to create such a unique gaming platform, researchers have teamed up with a well-known software
developer BetBuddy.The new version of BetBuddy platform has become one of the most advanced solutions in
the gaming industry. It keeps track of user’ s gambling behavior in casinos through using AI technologies.
Player’s behavior model is identified through using inference mechanisms and neural networks, as well as
random forest algorithms, indicating and signalizing the exact time when players reach problematic levels. This
mechanism enables online casino operators to decide either to block problematic accounts or to apply some
limitations to them, etc.
Features of the platforms
These days the most interesting solutions of practical implementation of AI technologies in gambling are
offered by two developers – BetBuddy and BtoBet.
Sabrina Soldà, the Head of Marketing at BtoBet, commented on the idea of AI implementation in the online
casino industry: “Today, due to highly competitive environment, the gaming market requires special tools and
intelligent platforms in order to adjust to the needs and expectations of players worldwide. Over the past year,
BtoBet solutions have caught the eye of many operators from different countries. For this very reason, we have
decided to expand the global market coverage.
BetBuddy developers and scientists from the University of London observed the following facts regarding
practical use of the AI technologies: “We have presented an innovative system with artificial intelligence aimed
at prediction of problems based on player’s behavior, as well as at displaying of potential gambling
addictions. It is solely based on mathematical algorithms. Integration of the platform into online casino
websites will help operators to monitor potential problematic players, as well as to improve their customer
service.”
The future belongs to AI
John McCarthy, who was the first one to introduce the concept of “artificial intelligence” back in 1956, said:
“As soon as it starts functioning, we no longer call it artificial.”
The mankind has often used the AI in various fields of its activity, without even thinking about it. It’s just an
everyday reality. Currently, the concept of the AI is applied practically everywhere, particularly in the areas
connected with computer technologies. As it is frequently claimed by various researchers, the potential power
of the AI is limitless and often exceeds human capacities.
Slotegrator’ s experts do not consider surprising the fact that high-tech and iGaming industries are actively
developing gradually integrating AI systems into modern online casino platforms. Taking into consideration the
latest trends, the future is all about the superpowers of Artificial Intelligence.

 

 

 

 

 

 

 

 

 

 

 

 

分類
人工智慧

這五種AI應用,即將影響未來博弈行業!

人工智慧深深影響新一代各產業的運作模式,博彩業即是其中與AI密不可分的行業。AI為博奕行業帶來許
多前所未有的解決方案,無論是線上產業不可或缺的大數據整合、預測行業趨勢提供有效資訊;或是資金
與數據的安全保護問題,亦能利用AI來防止網路攻擊。且看以下五大將影響博奕產業的AI技術! !
As the development of artificial intelligence grows increasingly more sophisticated and begins to become
embedded in a number of solutions, the potential for its use in the gambling sector is becoming more
apparent. There are a number of ways in which AI can be harnessed, with it being employed to bet professional
poker and Go players making headlines recently. In this article, five ways that AI could potentially be of use to
the gambling sector will be reviewed, with some examples of companies that are working in these areas.
1. KYC and AML Compliance
Compliance to regulations is a vital part of any business, both to protect the company and its customers.
Rigorous standards to identify illegal or at-risk consumers need to be pursued to avoid damage to a company
through fraudulent activity, breaking the law or even simply bad publicity. In order to comply with Know Your
Customer and Anti-Money Laundering regulations, accurate and reliable methods of identifying customers and
monitoring behaviour need to be employed. This is often a time consuming and costly task which is only
needed to catch the small percentage of people who are engaged with illicit activity or are putting their
livelihoods at risk due to an addiction. AI can offer a solution here to identify and monitor customers, with the
ability to flag up potential dangers.
A couple of examples of companies working to achieve this are Onfido and Trulico. London-based Onfido
enables remote background checks on customers and utilises a number of different verification means. The AI
and machine learning enables Onfido to strengthen its fraud detection further as more data is added. Trulico
gives the capability for automatic identity verification on a global scale. One goal of Truiloo is to bring
identification services to areas of the world where people have no other record of existence or struggle to
prove identity. GlobalGateway, the company’ s instant electronic identity verification service is designed
explicitly to enable businesses to comply with cross-border Anti-Money Laundering and Know Your Customer
regulation.
2. Prediction
One of the key factors in a business that is going to successfully keep up with developments in technology and
consumer trends is the ability to monitor consumer behaviour for trends in activity and suggestions of future
behaviour. By doing this, companies can design effective new products, tailor existing offerings and plan the
next steps to take in development. Al can enable this through its capacity to analyse and learn from large
amounts of data to provide accurate and up-to-date reports to a degree that a human equivalent would not be
able to achieve.
An example of AI being put to use in this manner is in Seldon. Another London-based company, the Seldon
platform has the ability to Predict media and e-commerce customer future actions on web, tablet and mobile
devices. Seldon analyses behaviour, social information, data from first and third party sources and any
contextual information in order to enhance product and content recommendations. Another example, Opera
Solutions uses AI to enable companies to draw predictive intelligence and conclusions from big data. It
identifies patterns to assist researchers in understanding developments on an industrial or global scale to allow
these developments to be taken advantage of at the earliest opportunity. Na
3. Simulation
AI could also be used to provide new experiences in gaming, perhaps opening up a new area for the gambling
industry target as a wider audience is drawn in.
Improbable is a company that aims to enable the building of ‘simulated worlds’ by combining different
servers and games engines to combine into one massive multiplayer experience. It has been described as trying
to create The Matrix’ . Besides creating virtual worlds, the platform would also allow for numerous
simulations to be run, potentially assisting many different types of company that want accurate prediction
models. The British start-up is valued at over $1 billion.
4. Security
Naturally, security is an area that companies require to be as strong as possible in order to protect their own
funds and data, but also to ensure that customers feel secure in using their services – leaving their privacy and
finances in safe hands. If customers feel their funds or information are at risk, they will choose other options for
their gambling needs. The theft of money and documentation, as well as bad publicity, are threats that every
company faces on a daily basis.
Based in Cambridge in the UK, Darktrace uses AI and machine learning to identify patterns with the ability to
detect and stop a cyber-attack before it occurs. This early-warning solution is an effective method that is
already being employed by companies such as BT and Virgin Trains. By preventing the attack before it even
happens, the risk of failure is decreased.
5. Automation
Not being a living system, AI can avoid many of the downsides that human operators have. Al does not get
tired or hungry, doesn’t make mistakes and has no need of sleep. Implementing AI can improve the efficiency,
speed and accuracy of previously manual tasks, freeing up manpower to be used in more important areas. AI
can particularly excel at this in areas where large amounts of repetitive tasks are being carried out.
One example of a company putting AI to this use is Tractable, which is building deep-learning tools to perform
tasks previously performed by experts relying on visual methods. The Tractable platform uses AI to process and
understand thousands of images in seconds with pin-point accuracy-outperforming human counterparts.
These are just some applications of AI that could apply to the gambling industry and more uses and more
companies developing Al solutions exist. In the future, the potential of AI will expand even further so
investigating and investing in Al capabilities will enable companies to prepare for future challenges and
opportunities.

 

 

 

 

 

 

 

 

 

分類
人工智慧

線上博彩商BtoBet 在ICE 2017全力展現AI新技術!

線上博彩商BtoBet
大膽研發AI
技術,被今年ICE Totaly Gaming鲁為以AI平台為發展核心的成功企業,更
受邀參加ICE VOX會議。本次會議與知名博彩商Microgaming合作,除了探討AI的發展未來,也表示十分
關注非洲市場,並著手將零售業務轉移至移動市場。
2
BtoBet has hailed this year’s ICE Totally Gaming as a success with the company
showcasing AI platforms and participating in ICE VOX conferences.
co Torded the company sexpecta
The iGaming operator, BtoBet, claims that ICE 2017 exceeded the company’s expectations in terms of the
interest in their product launches on the showfloor.
BtoBet were also chosen to take part in events across ICE as experts in the industry, including at ICE VOX and
the exclusive Microgaming yacht-conferences.
Discussing the balance of the three-day ICE show, BtoBeť s CEO, Kostandina Zafirovska, said: “Enthusiastic
attendees experienced trial runs of the Augmented Reality potential and took part in BtoBet’ s Virtual Assistant
Simone demonstration, participating in her show, interacting, taking pictures and having fun with her,” she
added. “Operators had the opportunity to see how BtoBeť s sophisticated technology can provide the
perfect integration between the A.1. platform and the behaviour of the player, suggesting the best games and
events at the ideal time to each player – through the Recommendation engine – and catch the trend of
Augmented Reality to improve and speed up new marketing strategies, providing players with the most
advanced and exciting gaming experience on the market.”
Zafirovska was also the technology speaker at the panel The Future of Trading: innovation at the Door’,
during BetMarkets session of the ICE Vox Conference – sponsored by Sportradar.
Commenting on the panel, she highlighted: “We had the occasion to show the urgent need of innovation and
the intelligent platform in the betting industry to manage risk and trade in an effective way, monitoring player
behaviour and preventing fraud with immediacy.”
In addition to this, BtoBet’ s chairman Alessandro Fried was selected as an expert speaker for the exclusive
Microgaming yacht-conference on the Sunborn London at Excel. The session, organized in partnership with
Microgaming, focused on the African Market, and showed operators ready-to-use tools, technology and
opportunities to differentiate their brand and expand their business from retail to mobile.

 

 

 

 

 

 

 

 

 

 

 

 

分類
人工智慧

機器自學 ! 青出於藍的AI人工智慧 !

近年來,AI人工智慧在各項益智遊戲中擊敗人類早已不是新聞。加拿大阿爾伯塔大學的科學家邁克爾·鮑靈
與他的AI撲克團隊,不斷地透過新的演算法及深度機器學習突破機器的規律性,成功促使其AI技術
DeepStack得以透過自學的方式,模仿人類大腦與習性,屢次在撲克遊戲中青出於藍,贏過人腦!
n
Two artificial intelligence (AI) programs have finally proven they “know when to hold’em, and when to fold
em,” recently beating human professional card players for the first time at the popular poker game of Texas
Hold’em. And this week the team behind one of those Als, known as DeepStack, has divulged some of the
secrets to its success-a triumph that could one day lead to Als that perform tasks ranging from from beefing
up airline security to simplifying business negotiations.
tyear one conquered Go
Als have long dominated games such as chess, and last year one conquered Go, but they have made relatively
lousy poker players. In DeepStack researchers have broken their poker losing streak by combining new
algorithms and deep machine learning, a form of computer science that in some ways mimics the human brain,
allowing machines to teach themselves.
“It’s a… a scalable approach to dealing with complex information) that could quickly make a very good
decision even better than people,” says Murray Campbell, a senior researcher at IBM in Armonk, New York,
and one of the creators of the chess-besting AI, Deep Blue.
Chess and Go have one important thing in common that let Als beat them first: They’re perfect information
games. That means both sides know exactly what the other is working with—a huge assist when designing an
AI player. Texas Hold’em is a different animal. In this version of poker, two or more players are randomly dealt
two face-down cards. At the introduction of each new set of public cards, players are asked to bet, hold, or
abandon the money at stake on the table. Because of the random nature of the game and two initial private
cards, players’ bets are predicated on guessing what their opponent might do. Unlike chess, where a winning
strategy can be deduced from the state of the board and all the opponent’ s potential moves, Hold ’em
requires what we commonly call intuition.
The aim of traditional game-playing Als is to calculate the possible results of a game as far as possible and then
rank the strategy options using a formula that searches data from other winning games. The downside to this
method is that in order to compress the available data, algorithms sometimes group together strategies that
don’t actually work, says Michael Bowling, a computer scientist at the University of Alberta in Edmonton,
Canada.
His team’ s poker AI, DeepStack, avoids abstracting data by only calculating ahead a few steps rather than an
entire game. The program continuously recalculates its algorithms as new information is acquired. When the AI
needs to act before the opponent makes a bet or holds and does not receive new information, deep learning
steps in. Neural networks, the systems that enact the knowledge acquired by deep learning, can help limit the
potential situations factored by the algorithms because they have been trained on the behavior in the game.
This makes the Al’ s reaction both faster and more accurate, Bowling says. In order to train DeepStack’s
neural networks, researchers required the program to solve more than 10 million randomly generated poker
game situations.
To test DeepStack, the researchers pitted it last year against a pool of 33 professional poker players selected by
the International Federation of Poker. Over the course of 4 weeks, the players challenged the program to
44,852 games of heads-up no-limit Texas Hold’em, a two-player version of the game in which participants
can bet as much money as they have. After using a formula to eliminate instances where luck, not strategy,
caused a win, researchers found that DeepStack’s final win rate was 486 milli-big-blinds per game. A milli-
big-blind is one-thousandth of the bet required to win a game. That’ s nearly 10 times that of what
professional poker players consider a sizable margin, the team reports this week in Science.
The team’s findings coincide with the very public success several weeks ago of Libratus, a poker AI designed
by researchers at Carnegie Mellon University in Pittsburgh, Pennsylvania. In a 20-day poker competition held in
Pittsburgh, Libratus bested four of the top-ranked human Texas Hold’ em players in the world over the course
of 120,000 hands. Both teams say their system’s superiority over humans is backed by statistically significant
findings. The main difference is that, because of its lack of deep learning, Libratus requires more computing
power for its algorithms and initially needs to solve to the end of the every time to create a strategy, Bowling
says. DeepStack can run on a laptop.
Though there’ s no clear consensus on which AI is the true poker champ—and no match between the two has
been arranged so far—both systems have are already being adapted to solve more complex real-world
problems in areas like security and negotiations. Bowling’ s team has studied how AI could more successfully
randomize ticket checks for honor-system public transit.
Researchers are also interested in the business implications of the technology. For example, an AI that can
understand imperfect information scenarios could help determine what the final sale price of a house would be
for a buyer before knowing the other bids, allowing that buyer to better plan on a mortgage. A system like
AlphaGo, the perfect information game-playing Aſ that defeated a Go world champion last year, couldn’t do
this because of the lack of limitations on the possible size and number of other bids.
ULTTa GarminG
Still, DeepStack is a few years away from truly being able to mimic complex human decision making, Bowling
says. The machine still has to learn how to more accurately handle scenarios where the rules of the game are
not known in advance, like versions of Texas Hold ’em that its neural networks haven’t been trained for, he
says.
Campbell agrees. “While poker is a step more complex than perfect information games,” he says, “it’s still
a long way to go to get to the messiness of the real world.”

 

 

 

 

 

 

 

 

分類
人工智慧

撲克玩家的地獄 ? AI強勢來襲 !

人工智慧當道,已成為市場顯學。舉凡任何與網路相關的產業,無不積極投入與開發AI的技術,其中博奕
遊戲產業也深受影響。匹茲堡超級計算中心的研究團隊以AI與四位世界級職業撲克選手展開人腦與電腦的
對決,並在競賽中赢得了170萬美元!本文為您詳細解析,A是如何擊敗四位世界頂尖撲克選手!
slavina Trocco hands of heads-ub. no-lim
“That was anticlimactic,” Jason Les said with a smirk, getting up from his seat. Unlike nearly everyone else in
Pittsburgh’s Rivers Casino, Les had just played his last few hands against an artificially intelligent opponent on
a computer screen. After his fellow players – Daniel McAulay next to him and Jimmy Chou and Dong Kim in an
office upstairs — eventually did the same, they started to commiserate. The consensus: That AI was one hell of a
player.
The four of them had spent the last 20 days playing 120,000 hands of heads-up, no-limit Texas Hold’em
against an artificial intelligence called Libratus created by researchers at Carnegie Mellon University. At stake: a
total pot of $200,000 and, on some level, the pride of the human race. A similar scene had unfolded two years
prior when Les, Kim and two other players decisively laid the smackdown on another Al called Claudico. The
players hoped to put on a repeat performance, finish up the event January 30th, and ride the rush of
endorphins until they got home and resumed their usual games of online poker.
The fight wasn’t even close. All told, Libratus won by more than 1.7 million (virtual) dollars, and — just like that
— the second Brains vs. Al competition came to a close. To understand what these players were up against and
what makes Libratus work, let’s go back to a time before all hope of victory was lost.
Men vs. machine
For the four men playing against Libratus, victory didn’t always seem impossible. The AI was in the lead from
the get-go, building an impressive streak of wins for the first three days. Then came the counter-attack. Day
four saw the gap narrow $40,000, and a string of successes on day six brought the humans to within $50,000 of
the lead.
“In the start here, we lost the first day,” Les explained. “Whatever — not a big deal. And then we were losing,
but then we fought back up to nearly equal. We were feeling really confident! We know how to play, we’re
going to be able to win.”
On the night after the sixth day of competition, the humans did what they did every other night: sift through
the Libratus hand data provided to them by CMU in hopes of devising a winning strategy. With spirits high after
a big day, they decided on a seemingly crazy strategy: three-betting on every hand that came along.
Three-betting, for the uninitiated, is poker slang for reraising on a hand. When you decide to play a hand in a
situation like this, paying the blinds is the first bet. If you’re confident in your cards, you raise — that’s the
second bet. Generally, when you reraise — the third bet-you’re pretty sure you’ve got the exchange in the
bag. Based on their understanding of Libratus’ play style, the humans thought they could knock if off balance
by playing this aggressively for a while. It backfired.
“We applied this crazy strategy we would never do online,” Kim explained. “Basically, we reraised all of our
hands. All of us went in, like, ‘Let’ s just try this, let’ s go crazy.”
“We had a reason to believe that specific size-three-bet was going to work well against the AI,” Les added.
“We just fired off all day doing that.”
Les and Kim concede that they just got unlucky, too, but either way: Libratus was unfazed by their plan and
started demolishing them. “It just kept improving every single day, and we started going backwards and
backwards,” Les said. In fairness, the humans weren’t playing with their usual setups. The four competitors
are almost exclusively online poker pros, and when duking it out at virtual tables at home, they always have
their HUDs handy. These heads-up displays are filled with stats and probabilities that help online players make
the best moves. Their absence here in Pittsburgh was noticeable.
“Without the HUD, without the numbers, you don’t know if you’re being paranoid or not,” Daniel McAulay
said, leaning back in his chair after winning a hand. “Is it folding less? We were never sure. We would always
say the same thing to each other. Just play it out until we get home and we’d see the sample of hands and
then we’ll change the plan. But that cost us a lot of money. A lot of money.”
Those losses would only continue to mount.
Building the beast
One of the men responsible for the players’ anguish can usually be found in his ninth-floor office, overlooking
Carnegie Mellon University’s snow-flecked quad. Professor Tuomas Sandholm might live a second life as a
startup entrepreneur, but he has spent years trying to perfect the algorithms that make Libratus such a potent
player. It wasn’t out of any particular love for the game — Sandholm admits he’s no poker pro — but he was
fascinated by the thought of complex computer systems that make decisions better than we can. That fixation
led him to co-create Claudico (the earlier AI that the humans trounced) with PHD student Noam Brown, and it
led the two of them to try again with Libratus.
To think of Libratus as just a poker-playing champ is to sorely underestimate it. Instead, Sandholm says, it’ sa
more general set of algorithms meant to tackle any information-imperfect situation. Confused? Don’t be.
Broadly speaking, the term just describes any situation in which two or more parties don’t have the same
information. Something unlike, say, chess, where the entirety of the game’ s world is splayed out on the board
in front of players. Those players can figure out exactly what’s going on and, assuming they have decent
memories, draw on their understanding of the events that led them there. This is a perfect information game.
No-limit Texas Hold’em is different. You don’t know which cards your opponent has, your opponent
doesn’t know which cards you have, and those minutes playing a hand to its conclusion are spent trying to
make the smartest moves possible with a shortage of intel. And unlike the limit variant, where there’s a cap on
how big your bets can be, no-limit gives you the freedom to bet whatever you want. There’s so much
information a person — or an AI — can infer about an opponent’ s strategy based on their bets that it sno
wonder researchers have been trying to crack the game.
“Heads-up, no-limit Texas Hold’ em poker has emerged as the leading benchmark for measuring the quality
of these general purpose algorithms in the AI community,” Sandholm told me.
With that in mind, Sandholm and Brown jointly built Libratus from three major components. The first is an
algorithm that devises overall strategies based on Nash equilibria. In other words, Libratus spent a total of 15
million computing hours chewing on the rules of the game before the competition, finding rational ways to act
when both players are making the best possible moves with the information available. Thanks to a new logic
model developed by the two researchers to minimize Libratus’ “regret,” the AI could solve larger
abstractions of the game faster and with higher accuracy than before.
The second is what Sandholm calls the end-game solver. This is the part that players actually faced during their
20 days of combat. Unsurprisingly, too, this is where Sandholm says most innovative breakthroughs have
happened. Essentially, this allowed Libratus to cook up an approach based on the first two cards it was dealt,
and modify that approach based on its opponent’ s actions and the river and flop that are dealt. Sandholm
says Libratus was also designed to keep tabs on how safe its options are. Let’s say a human player screws up
and loses $372. That money is viewed as a gift of sorts, so the AI can freely lose up to $372 and still remain
ahead.
ULTTE Garming
“That gives us more flexibility for optimizing our strategies while still being safe,” Sandholm explained.
We’ II get to the last key component a little later. In any case, the sheer number of complex calculations meant
Libratus couldn’t run on the desktop in Sandholm’ s office. If nothing else, the human players can take solace
in the fact that it took a supercomputer and millions of computing hours to beat them. If you thought Gowas
tough to wrap your head around, consider the complexity of no-limit Texas Hold’em: When you’re dealt into
a game, the hands you’re dealt and the communal cards that appear are one possibility of 10^160.
“That’s one followed by 160 zeroes,” said Sandholm. “That’s more than the number of atoms in the
universe. You cannot just brute-force your way through it.” Still, it takes some degree of brute force to build as
close to optimal a strategy as possible. That’s where “Bridges” comes in.
If Libratus is the brain of the operation, Bridges — a supercomputer made of hundreds of nodes in the
basement of the Pittsburgh Supercomputing Center – is most definitely the brawn.
“Libratus is running on about 600 nodes at Bridges, out of 846 total compute nodes,” said Nick Nystrom,
senior director of research at the Pittsburgh Supercomputing Center. Most of those 800+ nodes have two
CPUs, each with 28 computing cores and 128GB of RAM. Forty-eight of those nodes have two state-of-the-art
GPUs, and still others were loaded with even more power: NVIDIA’ s Tesla-series K80 and P100 GPUs.
༧.༠༠
There’s more: 42 of those nodes have 3TB of RAM each, and a very special four nodes have a whopping 12TB
of RAM. That’s some serious firepower, but all those nodes were ingeniously woven together to maximize
data bandwidth and minimize latency. It’s just as well, considering the amount of data involved: Libratus was
using up to 2.6 petabytes of storage during the competition.
When not being used to best humans at card games, Bridges was being used for around 650 projects by more
than 2,500 people. Think of Bridges as a supercomputer for hire: Researchers from around the country are using
it to gain insight into arcane subjects like genomics, genome sequence assemblies and other kinds of machine-
learning.
The beauty of Bridges, according to Nystrom, is that those researchers don’t need to be supercomputer buffs.
“It’ sa very cloud-like model letting people who are not programmers, not computer scientists, not
supercomputer users make use of a supercomputer without necessarily even knowing it.” That’ s what
happened with Libratus, and everything seemed to be working perfectly.
ULTS 2 Garmin
GATING
Game theory
After the humans’ gutsy attack plan failed, Libratus spent the rest of the competition inflating its virtual
winnings. When the game lurched into its third week, the AI was up by a cool $750,000. Victory was assured, but
the humans were feeling worn out. When I chatted with Kim and Les in their hotel bar after the penultimate
day’s play, the mood was understandably somber.
“Yesterday, I think, I played really bad,” Kim said, rubbing his eyes. “I was pretty upset, and I made a lot of
big mistakes. I was pretty frustrated. Today, I cut that deficit in half, but it’s still probably unlike for me to
win.” At this point, with so little time left and such a large gap to close, their plan was to blitz through the
remaining hands and complete the task in front of them.
For these world-class players, beating Libratus had gone from being a real possibility to a pipe dream in just a
matter of days. It was obvious that the AI was getting better at the game over time, sometimes by leaps and
bounds that left Les, Kim, McAulay and Chou flummoxed. It wasn’t long before the pet theories began to
surface. Some thought Libratus might have been playing completely differently against each of them, and
others suspected the AI was adapting to their play styles while they were playing. They were wrong.
As it turned out, they weren’t the only ones looking back at the past day’ s events to concoct a game plan for
the days to come. Every night, after the players had retreated to their hotel rooms to strategize, the basement
of the Supercomputing Center continued to thrum. Libratus was busy. Many of us watching the events unfold
assumed the AI was spending its compute cycles figuring out ways to counter the players’ individual play
styles and fight back, but Professor Sandholm was quick to rebut that idea. Libratus isn’t designed to find
better ways to attack its opponents; it’ s designed to constantly fortify its defenses. Remember those major
Libratus components I mentioned? This is the last, and perhaps most important, one.
“All the time in the background, the algorithm looks at what holes the opponents have found in our strategy
and how often they have played those,” Sandholm told me. “It will prioritize the holes and then compute
better strategies for those parts, and we have a way of automatically gluing those fixes into the base strategy.”
If the humans leaned on a particular strategy — like their constant three-bets — Libratus could theoretically
take some big losses. The reason those attacks never ended in sustained victory is because Libratus was quietly
patching those holes by using the supercomputer in the background. The Great Wall of Libratus was only one
reason the AI managed to pull so far ahead. Sandholm refers to Libratus as a “balanced” player that uses
randomized actions to remain inscrutable to human competitors. More interesting, though, is how good
Libratus was at finding rare edge cases in which seemingly bad moves were actually excellent ones.
“It plays these weird bet sizes that are typically considered really bad moves,” Sandholm explained. These
include tiny underbets, like 10 percent of the pot, or huge overbets, like 20 times the pot. Donk betting, limping
— all sorts of strategies that are, according to the poker books and folk wisdom, bad strategies.” To the
players’ shock and dismay, those “bad strategies” worked all too well.
Poker and beyond
On the afternoon of January 30th, Libratus officially won the second Brains vs AI competition. The final margin
of victory: $1,766,250. Each of the players divvied up their $200,000 spoils (Dong kim lost the least amount of
money to Libratus, earning about $75,000 for his efforts), fielded questions from reporters and eventually left to
decompress. Not much had gone their way over the past 20 days, but they just might have contributed to a
more thoughtful, Al-driven future without even realizing it.
Through Libratus, Sandholm had proved algorithms could make better, more-nuanced decisions than humans
in one specific realm. But remember: Libratus and systems like it are general-purpose intelligences, and
Sandholm sees plenty of potential applications. As an entrepreneur and negotiation buff, he’ s enthusiastic
about algorithms like Libratus being used for bargaining and auctions.
“When the FCC auctions spectrum licenses, they sell tens of billions of dollars of spectrum per auction, yet
nobody knows even one rational way of bidding,” he said. “Wouldn’t it be nice if you had some AI
support?”
But there are bigger problems to tackle – ones that could affect all of us more directly. Sandholm pointed to
developments in cybersecurity, military settings and finance. And, of course, there’s medicine.
“In a new project, we’re steering evolution and biological adaptation to battle viral and bacterial infections,”
he said. “Think of the infection as the opponent and you’re taking sequential actions and measurements just
like in a game.” Sandholm also pointed out that such algorithms could even be used to more helpfully
manage diseases like cancer, both by optimizing the use of existing treatment methods and maybe even
developing new ones.
Jason, Dong, Daniel and Jimmy might have lost this prolonged poker showdown, but what Sandholm, Brown
and their contemporaries have learned in the process could lead to some big wins for humanity.