Baidu Wu Enda: Computer hardware capabilities will promote a new wave of artificial intelligence development

Baidu chief scientist Andrew Ng said recently that the enhancement of computer hardware capabilities will provide the impetus for the further development of artificial intelligence (AI). Therefore, the current round of AI heat will not be as short-lived as before, and "AI Winter" may not appear again.

Baidu Wu Enda: Computer hardware capabilities will promote a new wave of artificial intelligence development

At present, the development of artificial intelligence is in full swing. Researchers are releasing the latest progress in the field of AI at a dizzying speed. Many companies have established their own professional AI team as quickly as possible. Given the history of AI development in the past, one cannot help but ask: Can this glorious scene continue?

Wu Enda, chief scientist, machine learning and AI expert at Baidu Research Institute, said that in the foreseeable future, improvements in computer processor design will bring about performance improvements and breakthroughs, which will provide power for the further development of AI. He said: "A number of hardware vendors have shared their product roadmaps with us. To be sure, they are trustworthy, and we will have more computing power and faster networks in the next few years."

AI has experienced rapid development and over-promotion in the past, but it has cooled down quickly, and the amount of investment and people's interest has plummeted. It is called “AI Winter”. The first winter of AI appeared in the 1970s, when the technology slowed down and the government invested. Another cold winter occurred in the 1980s because AI failed to achieve the expected commercial impact.

Next, AI has seen this current boom, mainly due to the rapid development of machine learning technology. Nowadays, major companies have set up AI research teams, and hired AI researchers to conduct basic research with high salaries, hoping to achieve major business breakthroughs.

Thanks to the development of the “deep learning” system, AI has made remarkable progress in recent years. Beginning a few years ago, researchers have found that "labeled examples" can be used to train very large or very deep neural networks that allow them to identify things with near-human accuracy. Today, the technology is making amazing advances in areas such as image and speech recognition.

Wu Enda said that these systems will only become stronger. Not only can the accuracy of current deep learning tools be improved, but it can also be applied to other new areas such as analytics and generative languages. In addition, advances in hardware will also power the development of emerging AI technologies.

Wu Enda said: "If the hardware computing power is increased by 10 times, you can carry out many experiments I want to do." For example, more powerful computing power will allow us to build a single algorithm to perform various image-related tasks, instead of the current situation, Use a bunch of different image processing algorithms.

On Wednesday, the 2016 Neurological Information Processing System Deep Learning Conference was held in Barcelona, ​​bringing together AI experts from around the world. A few years ago, the conference was attended by only a few hundred people, and this year's participation exceeded 6,000, highlighting the strong interest in AI.

Wu Enda said: "There must be some exaggeration here, but I believe that this time the AI ​​fever has real value and will not be as short-lived as in previous years."

Richard Socher, chief scientist, machine learning and linguist at Salesforce, said that the vast amount of data available, as well as advances in machine learning algorithms, will drive AI forward.

Socher also said that the current machine learning technology is limited to some applications launched by a few large enterprises in the consumer market, and the expansion of the technology into a wider field will bring greater influence. He said: "If Salesforce's 150,000 corporate customers increase their operational efficiency by 1% through machine learning technology, then the US's GDP will increase significantly."

Soher believes that the application of machine learning in all walks of life will ensure that AI maintains sufficient market interest. He said: "Even if there is, the winter of AI will not be so cold."

30W Medical Power Supply

30W Medical Power Supply,Dr Adapter,B Ultrasound Power Supply,B Ultrasound Adapter

Shenzhen Longxc Power Supply Co., Ltd , https://www.longxcpower.com

This entry was posted in on