新港报告——人工智能与科技革命系列第二场：人工智能与大脑智能
讲座名称：
新港报告——人工智能与科技革命系列第二场：人工智能与大脑智能
讲座时间：
2020-10-17
讲座人：
Colin Blakemore
形式：
校区：
兴庆校区
实践学分：
讲座内容：
新港报告——人工智能与科技革命系列第二场
讲座题目：人工智能与大脑智能
讲座时间：10月17日下午15：00
讲座地点：涵英楼新港报告厅
讲座人：Colin Blakemore
讲座内容：人工智能与大脑智能
AI (Artificial Intelligence) and BI (Brain Intelligence)
From AlphaGo to driverless cars, AI has made remarkable progress in the past decade. This is partly because of faster processing speed but it is largely due to the use of massive machine learning, based on neural network architecture. Superficially, this approach to computing has similarities to the structure and function of the brain, which has huge number of neurons and very rich connectivity between them, with connections (synapses) that can change their transmission strength according to simple rules. The current slogan, ‘Brain-Inspired AI’ – seen everywhere in China – expresses the hope that the brain might hold other secrets of neural computation that could be copied in future generations of AI. This might be true, but it is important to understand that our knowledge of the brain is still very elementary and the probabilistic, analog nature of brain computation is, in many ways, fundamentally different from the principles of machine computation.In fact, it is possible that developments in AI will tell us more about how the brain works, than brain research tells us about AI.
从AlphaGo到无人驾驶汽车，人工智能在过去的十年里取得了显著的进步。这部分是因为处理速度更快，但很大程度上是由于使用了基于神经网络架构的大规模机器学习。从表面上看，这种计算方法与大脑的结构和功能有相似之处，大脑有大量的神经元，它们之间有非常丰富的连通性，连接（突触）可以根据简单的规则改变它们的传输强度。目前在中国随处可见的“大脑启发型人工智能”的口号表达了一种希望，即大脑可能掌握神经计算的其他秘密，这些秘密可以在未来几代人工智能中复制。这也许是真的，但重要的是要明白，我们对大脑方面的知识仍然是非常基础的，而且大脑计算的概率性、模拟性在许多方面与机器计算的原理根本不同。事实上，相比于大脑研究告诉我们的人工智能，人工智能的发展有可能更多地告诉我们大脑是如何工作的。
相关视频