AI visionary and proponent Andrew Ng wears many hats. Of late, the former Googler (led the Google Brain project) and co-founder of Coursera, the popular online education portal has been grabbing headlines for his high profile exit from Baidu, China’s online search giant.
Announcing his departure in a post, Ng detailed Baidu’s AI strategy, wherein he played the key role of being the principal architect, charting the incredible rise of AI within the company. “My team birthed one new business unit per year and each of the last two years — autonomous driving and the DuerOS Conversational Computing platform,” Ng shared in the post. The prolific computer scientist further added, “We are also incubating additional promising technologies, such as face-recognition (used in turnstiles that open automatically when an authorized person approaches), Melody (an AI-powered conversational bot for healthcare) and several more”.
At Baidu, China’s Google, Ng who held the mantle of VP and chief scientist led a team of AI group from the Sunnyvale campus and transformed it into one of the one of the few companies with world-class expertise in every major AI area such as speech recognition, NLP, computer vision and machine learning. For Ng, AI is the new electricity: Just as electricity transformed many industries roughly a hundred years back, AI will also now change practically every major sector from healthcare, transportation, to entertainment enriching the lives of countless people. I am more excited than ever about where AI can be.
His seminal work in Deep Learning is revolutionizing industries
In an earlier talk, Ng revealed that ever since he was a high school student, as a 17-year-old, he had been fascinated with AI. During his early years at Stanford University, he saw from close quarters “how really complicated this field was going”. “I thought, others are writing really complex programs but we’re Stanford University, we can write even more complex programs”.
True to his statement, that is what happened for some years but it never really worked. At one point, Ng thought “it will be very hard to make progress in AI”. Somewhere around 1998, the machine learning researcher came across an idea that changed his life and gave him hope that progress in AI was within reach. “So, rather than all of us going off and writing increasingly complex problems, we looked at the human brain for inspiration and there is a fascinating hypothesis. This is not a proven fact, but a hypothesis that the way brain does perception is not by having 50, 100 or 1000 really complicated programs. Most of the perception in the brain may be due to one learning algorithm,” shared Ng.
Google Brain – marks Ng’s shift from academic to industry
What we now know as the Google Brain project founded in 2011, was based on the “fascinating experiment wherein all of human learning is based on neurons”. “What a large community of researchers did was work on artificial neural networks which is software simulating how the brain might work. Using these artificial simulations in the brain we have been able to tackle a number of things,” explained Ng, talking about the groundbreaking research that led to the founding of Google Brain.
His seminal work led to massive improvements across the spectrum – search, voice and image search. At Google Brain, Ng’s team of researchers leveraged massive amount of data to solve problems that weren’t easy to tackle by machines, especially in computer vision such as object classification and in NLP, producing natural language captions for images.
Interestingly, out of the many pivotal projects carried out by Google Brain, the one that made most headlines was where Google’s Artificial Brain learnt to find cats in Youtube videos. His Google Brain team got 16,000 computers to train themselves to find images of cats in videos, that too by making them watch YouTube videos. The findings were extremely useful in the development of image and speech recognition software and also translation.
Birthing MOOC, that paved the way for Coursera – popular ed tech startup
The prominent Stanford academic and the Director of the Stanford Artificial Intelligence Lab, Ng pioneered Stanford’s first attempt at free, online distributed education, called Stanford Engineering Everywhere (SEE) in 2008. In 2011, the launch of three successful MOOCs led to the inception of Coursera that has become a platform that works with existing universities to offer blended learning experience.
Why is the industry betting big on Deep Learning?
According to Ng, Deep learning is one of the technologies that many tech companies are heavily investing in. “It is driving a large amount of economic value and there is a growing impact on enterprises and one of the main drivers of AI,” he said. Decoding the buzzword, he divulged, Deep Learning is a broad term shared by many researchers and it is done by building very large scale simulations of the brain. In fact, some of the technology used today is probably powered by deep learning, he added.
In fact, it’s not just tech heavyweights such as Intel, IBM, Microsoft that are making a Deep Learning push by stacking up their AI portfolio, startups such as Affectiva, Nervana Systems (acquired by Intel) and Deep Genomics are also leading the charge in AI.
Will Ng move to Didi Chuxing, China’s top ride-sharing company?
Speculations are rife that Ng’s resignation was motivated by China’s ride sharing company Didi Chuxing, setting up a lab in Mountain View. Ng hinted about scripting a new chapter in AI in his post, and if news reports are anything to go by, Didi Chuxing recently formed a Mountain View-based lab for promoting AI and self-driving car technology. The company recently snagged up high profile appointment, Uber employee Charlie Miller came onboard and Ng could be the next line.
Try deep learning using MATLAB