新奇事件簿 OpenAI發(fā)布增強(qiáng)學(xué)習(xí)算法測(cè)試平臺(tái)

00:00

00:00

MP3下載 文本下載

https://online2.tingclass.net/lesson/shi0529/10000/10315/452.mp3

https://image.tingclass.net/statics/js/2012

A group of business people has opened a kind of gym for robots. The group is headed by Elon Musk, who is the boss of the electric car company Tesla Motors and the space tourism company, SpaceX. Mr Musk also founded the online payments website PayPal. His latest project is called OpenAI Gym. It is a platform for researchers and code writers to test their latest work. It is an open platform, which means people are free to test their ideas for artificial intelligence (A.I.) on it. In particular, researchers can test and share their algorithms for A.I. Algorithms are special sets of rules in a computer program that can solve problems and quickly deal with large amounts of information.

商界人士為機(jī)器人開設(shè)“健身房”。發(fā)起人是埃隆·馬斯克，埃隆是電動(dòng)汽車特斯拉總裁，同時(shí)也是太空旅行SpaceX總裁。他還創(chuàng)立了網(wǎng)上支付PayPal。最新項(xiàng)目名叫“OpenAI Gym”。該平臺(tái)可供研究人員和代碼編寫者測(cè)試新成果。平臺(tái)向社會(huì)公開，民眾可免費(fèi)測(cè)試最新人工智能成果。特別要提到的是，研究人員可分享演算。演算法是計(jì)算機(jī)程序中獨(dú)有的規(guī)則，它可以解決問題，迅速處理大規(guī)模信息。

Mr Musk wrote about the OpenAI Gym project and why his team decided to put it online for free. He said: "Our goal is to advance digital intelligence in the way that is most likely to benefit humanity as a whole, without the need to generate financial return." The OpenAI Gym follows the idea of giving rewards for successful learning similar to giving a dog a treat when it learns something new. This is called reinforcement learning. If an algorithm does well in the gym, it gets a reward. If the algorithm fails, it gets no reward and it has to try something different. The aim of the gym is to develop an algorithm that can multi-task do many things at the same time rather than be good at doing just one thing.

馬斯克對(duì)開設(shè)“OpenAI Gym”以及設(shè)立免費(fèi)網(wǎng)絡(luò)平臺(tái)進(jìn)行了描述。他稱：“我們的目的就是要加快數(shù)字化智能發(fā)展，最大可能造福全人類。無需產(chǎn)生經(jīng)濟(jì)回報(bào)。”類似小狗習(xí)得新事物獲得獎(jiǎng)勵(lì)，學(xué)習(xí)效果顯著也會(huì)得到獎(jiǎng)勵(lì)。我們稱之為增強(qiáng)學(xué)習(xí)算法。若演算法在“Gym”表現(xiàn)優(yōu)異，它會(huì)獲得獎(jiǎng)勵(lì)。若演算失敗，不會(huì)獲得獎(jiǎng)勵(lì)，但要嘗試新算法。“Gym”目的旨在發(fā)展多任務(wù)演算法，可同時(shí)兼做許多事情，而不是只擅長(zhǎng)某一領(lǐng)域。