[新聞] Google發現新算法看一次就可辨識新事物消失

看板Gossiping作者prime2477時間7年前 (2016/11/05 04:53)推噓15(16推 1噓 4→)

留言21則, 21人參與討論串1/1

1.媒體來源: MIT Technology Review - Robotics 2.完整新聞標題: Machines Can Now Recognize Something After Seeing It Once 現在只需看過一次，機器就可以辨識新事物 Algorithms usually need thousands of examples to learn something. Researchers at Google DeepMind found a way around that. 演算法經常需要成千上萬個樣本才能學習事物。Google DeepMind 研究員發現了一種方法可以繞過去。 3.完整新聞內文: by Will Knight November 3, 2016 Most of us can recognize an object after seeing it once or twice. But the algorithms that power computer vision and voice recognition need thousands of examples to become familiar with each new image or word. 大多數人只需看過一兩次就可以辨識一個物體。但是演算法驅動的電腦視覺與聲音辨識需要上千個樣本才能熟悉一個新的圖片或單詞。 Researchers at Google DeepMind now have a way around this. They made a few clever tweaks to a deep-learning algorithm that allows it to recognize objects in images and other things from a single example—something known as "one-shot learning." The team demonstrated the trick on a large database of tagged images, as well as on handwriting and language. https://arxiv.org/abs/1606.04080 現在 Google DeepMind 的研究員發現一個方法可以繞過去。他們對深度學習演算法做了一些聰明的小改變，使得只需一個樣本就可以辨認事物 — 號稱 "one-shot learning"。研究團隊使用大量數據演示了他們的方法。 The best algorithms can recognize things reliably, but their need for data makes building them time-consuming and expensive. An algorithm trained to spot cars on the road, for instance, needs to ingest many thousands of examples to work reliably in a driverless car. Gathering so much data is often impractical—a robot that needs to navigate an unfamiliar home, for instance, can’t spend countless hours wandering around learning. 目前最好的演算法可以穩定的辨識事物，但是它們對數據的需求，使得構建它們很耗時和昂貴。例如，要在無人駕駛汽車上訓練一個演算法辨認出路上的車子，需要輸入上千個樣本才能穩定工作。要收集如此多的資料是不切實際的 — 例如一個需要在陌生的家庭中導航的機器人，不可能讓它在家中花費無數時間徘徊練習。 Oriol Vinyals, a research scientist at Google DeepMind, a U.K.-based subsidiary of Alphabet that’s focused on artificial intelligence, added a memory component to a deep-learning system—a type of large neural network that’s trained to recognize things by adjusting the sensitivity of many layers of interconnected components roughly analogous to the neurons in a brain. Such systems need to see lots of images to fine-tune the connections between these virtual neurons. 專攻人工智慧的 Alphabet 子公司，Google DeepMind 的科學家 Oriol Vinyals，加了一個記憶元件到 deep-learning system — 一種大型的神經網路，以粗略模仿腦內神經元的元件構成，藉由調整多層相互連結的元件的靈敏度來辨識物體。這樣的系統需要看大量的圖片才能調整好虛擬神經元間的連結。 The team demonstrated the capabilities of the system on a database of labeled photographs called ImageNet. The software still needs to analyze several hundred categories of images, but after that it can learn to recognize new objects—say, a dog—from just one picture. It effectively learns to recognize the characteristics in images that make them unique. The algorithm was able to recognize images of dogs with an accuracy close to that of a conventional data-hungry system after seeing just one example. http://image-net.org/ 研究團隊是用了叫 ImageNet 的標記相片資料庫來演示系統的能力。該軟體仍然需要分析數百種類的圖片，但之後只需一張圖片，便可學習辨認新事物。它可以有效的學習辨認出相片中獨一無二的特徵。該演算法只需一個樣本，辨認狗照片的精確度就已經很接近傳統嗜資料如命的演算法。 Vinyals says the work could be especially useful if it could quickly recognize the meaning of a new word. This could be important for Google, Vinyals says, since it could allow a system to quickly learn the meaning of a new search term. Vinyals 表示如果這個研究成果可以用來快速的辨識出新詞的意義，那會非常有用。這對 Google 非常重要，因為它可讓系統很快地學習新的搜尋詞組的意義。 Others have developed one-shot learning systems, but these are usually not compatible with deep-learning systems. An academic project last year used probabilistic programming techniques to enable this kind of very efficient learning (see "This Algorithm Learns Tasks As Fast As We Do"). https://goo.gl/4hRWmY 也有其他人發展出 one-shot learning 系統，但是通常與 deep-learning 系統不相容。去年一個學術計畫以 probabilistic programming 技術使這種非常有效率的學習法成為可能。 But deep-learning systems are becoming more capable, especially with the addition of memory mechanisms. Another group at Google DeepMind recently developed a network with a flexible kind of memory, making it capable of performing simple reasoning tasks—for example, learning how to navigate a subway system after analyzing several much simpler network diagrams (see "What Happens When You Give a Computer a Working Memory?"). https://goo.gl/IMbG1q 但是 deep-learning 系統更有能力，尤其是加上了記憶機制。最近 Google DeepMind 的其他團隊發展出一個有彈性記憶的網路，它可以執行簡單的推理任務 — 例如，在分析數個較簡單的地鐵網路圖後，學習如何導航地鐵系統。 "I think this is a very interesting approach, providing a novel way of doing one-shot learning on such large-scale data sets," says Sang Wan Lee, who leads the Laboratory for Brain and Machine Intelligence at the Korean Advanced Institute for Science and Technology in Daejeon, South Korea. "This is a technical contribution to the AI community, which is something that computer vision researchers might fully appreciate." 韓國科學技術院，腦與機器智慧實驗室 Sang Wan Lee 表示 "我認為這是一個非常有趣的方法，它提供了一個新奇的手段在巨量資料上做 one-shot learning"。"這是個對AI社群的技術貢獻，電腦視覺的研究者可能會完全理解"。 Others are more skeptical about its usefulness, given how different it still is from human learning. For one thing, says Sam Gershman, an assistant professor in Harvard's Department for Brain Science, humans generally learn by understanding the components that make up an image, which may require some real-world, or commonsense, knowledge. For example, "a Segway might look very different from a bicycle or motorcycle, but it can be composed from the same parts." 其他人質疑它的實用性，並指出它與人類學習的差異。哈佛腦科學系助理教授 Sam Gershman 表示，人們通常藉由了解圖像的構成元件來學習，這需要一些現實生活的常識或知識。例如，Segway 的外觀與腳踏車或摩托車相比非常不一樣，但是它們都由相同的元件構成。 According to both Gershman and Wan Lee, it will be some time yet before machines match human learning. "We still remain far from revealing humans’ secret of performing one-shot learning," Wan Lee says, "but this proposal clearly poses new challenges that merit further study." 根據 Gershman 和 Wan Lee 的說法，在機器可以與人類學習法競爭之前，還有很長的一段路要走。Wan Lee 表示，"我們依然停留在距離揭祕人腦運用 one-shot learning 很遠的地方。但是這個計畫提出了值得更多研究的新挑戰"。 4.完整新聞連結 (或短網址): https://goo.gl/ywNfG8 5.備註: one-shot learning -- ※ 發信站: 批踢踢實業坊(ptt.cc), 來自: 36.227.0.122 ※ 文章網址: https://www.ptt.cc/bbs/Gossiping/M.1478292814.A.DFD.html

推

slimak

11/05 04:55, , 1^F

11/05 04:55, 1^F

推

yuanyo1222

11/05 04:56, , 2^F

11/05 04:56, 2^F

推

tentenlee

11/05 04:58, , 3^F

11/05 04:58, 3^F