不知道如何写作文章，推荐国内最强的AI写作工具，帮你写爆款文案

有多个小组研究该语言 (There are into the )

side of , among them team at IBM. Led by Gray, VP of IBM AI based in the ’s lab near New York, the are on in AI for . “ AI is not cool ; deep is cool. So we’re in a — or you can look at it as we’re ahead of the game,” Gray. “We think we’re ahead of the game.”

在机器智能方面，其中包括IBM的另一个团队。 由位于纽约州约克镇实验室的IBM AI科学副总裁 Gray领导，研究人员依靠统计AI的最新进展进行自然语言处理。 “经典的AI不再酷了； 深度学习很酷。 所以我们绝对是少数派，或者您可以在我们领先于游戏的同时看看它，” Gray笑着说。 “我们认为我们领先于游戏。”

His aim is to move from pure black box net to that can be as logic-like — but not the from . “You can’t rely on a bunch of to write down all the in the world,” says Gray. “, we’re going to learn that , to it from text.”

他的目标是逐步从纯黑匣子神经网络模型过渡到可以理解为类似于逻辑的知识的模型，但不一定是从人类身上获得的知识。 格雷说：“您不能依靠一群人来写下世界上所有的知识。” “相反，我们将学习该知识，以自动从文本中获取知识。”

For the past few years, Gray and his team have been using so- , a into a logic-like — the words to . “Take the ‘Mary had a lamb’ — we will the word Mary, and map it to the of a a graph, the use of other rich , such as the fact that a is a kind of , which is a kind of thing, and so on. This us to apply sense that the can use to more tasks,” says Gray.

在过去的几年中，Gray和他的团队一直在使用所谓的语义解析，将自然语言的句子翻译成类似逻辑的句子-将单词映射到明确的符号概念。 “采用“玛丽有只小羊羔”这一短语-我们将识别“玛丽”一词，并将其映射到知识图中的一个人的概念智能写作机器人，从而允许使用其他丰富的信息，例如一个人是一种哺乳动物，这是一种生物，等等。 这使我们能够应用机器可以用来执行更一般任务的常识。” Gray说。

part of the will that . “The of using a model which has a logic-like form is that you can then to get the to more ,” says Gray. “This is a path to true .”

研究程序的另一部分将自动获取该知识。 “使用具有类似逻辑形式的模型的优势在于，您可以执行推理来获得更复杂问题的答案，”格雷说。 “这是通往真正自然语言理解的可能途径。”

team at the MIT-IBM AI Lab is also in and . The an the Neuro- , where an AI with two about in . One a table with of the such as color, and size. The other one is on - pairs, such as “What’s the color of the cube?” — “Red.” That net then each into a AI that the table to get an .

MIT-IBM AI Lab的另一个团队也对将视觉和语言结合感兴趣。 研究人员开发了一种称为神经符号概念学习器的算法，其中具有两个神经网络的AI可以回答有关图像中对象的问题。 一个网络创建一个具有对象特征(例如颜色，位置和大小)的表。 另一个在问题-答案对上接受了培训，例如“立方体的颜色是什么？” —“红色”。 然后，该神经网络将每个问题转换成一个符号AI程序，该程序引用该表以获得答案。

That’s — the of a of light our and the data into the brain, which then it into we can in . , the have been able to relax the of that the has to have. First, it had to know the kinds of that were there, their and sizes. Then the knew that there was color but it didn’t know that blue or red are — it had to it out from and learn how it was tied to . And , the didn’t even know that color was a , it had to it out and learn what the color to. “They’ve been on this where the is given less and less and it has to learn — , sense,” says Cox.

那就是感知力-相当于光的光子撞击我们的视网膜并将视觉数据流到大脑，然后将其转换为我们可以用语言描述的东西。 至关重要的是可以帮你写爆款文案的AI系统，研究人员已经能够逐渐放松系统必须具备的先天知识。 首先，它必须知道那里的各种物体，它们的颜色和大小。 然后智能写作机器人，系统知道有一种叫做颜色的东西，但它不知道蓝色或红色是颜色-它必须从上下文中找出颜色，然后隐式地了解它与语言的联系。 最后，系统甚至不知道颜色是一个概念，它必须弄清楚颜色并了解颜色对应的含义。 Cox说：“他们一直在这个有趣的过程中，系统的使用越来越少，它必须学习—基本上是在发展常识。”

手机浏览，点击图片保存二维码到相册，然后打开微信扫一扫选择本二维码图片就可以进入，电脑端微信“扫一扫”二维码，进入找聊天搭子平台，里面有找饭搭子、找对象、找陪伴服务等等