AsianScientist (Jan. 25, 2018) – When computers first appeared on the scene in the 1950s, they were massive and expensive. Few could afford the processing power these machines offered, prompting computer scientist John McCarthy to develop a time-sharing system. Little did he realize that by allowing multiple users to concurrently tap on a single centralized piece of hardware, he was inadvertently sowing the seeds of what would later come to be known as cloud computing.
Since then, cloud computing has grown into a verdant ecosystem of software-, platform- and infrastructure-as-a-service computing models hosted on the internet. Leveraging the cloud, individuals and companies around the world can store and process data at scale, as well as develop custom applications in a flexible, cost-efficient manner, without themselves owning costly data centers. On the back of strong demand, the global market for cloud computing services is forecasted to soar to US$236 billion by 2020, with a compound annual growth rate of 22 percent.
As the cloud continues to expand, it has begun to converge with another revolutionary frontier in computing—artificial intelligence (AI). The smart algorithms, artificial neural networks and machine learning methods used in the development of AI have a voracious appetite for data. They also require substantial processing power to crunch larger and more complex datasets. The cloud, with its scalability and burgeoning stockpile of digital records, is thus a natural companion to AI.
“Cloud empowers AI technology by providing a powerful computational infrastructure, which will in turn help speed up the development of AI,” Dr. Min Wanli, chief AI scientist at Alibaba Cloud, told Asian Scientist Magazine. “By exploiting the massive data processing power of the cloud, AI researchers can develop more sophisticated algorithms on significantly larger training datasets to achieve more breakthroughs.”
High hopes
In the short span of its existence, AI backed by cloud computing has already been hailed as a driver of the next industrial revolution. Like fire to fuel, cloud-AI promises to unlock the value of data, blazing a path towards more efficient and productive enterprises and societies.
“Previously, the focus was simply on the visualization of Big Data, but this is not enough. What you want is actionable insights, and in real time,” said Min, noting that this is precisely what cloud-AI offers.
He cited the example of the most recent November 11 Singles’ Day shopping frenzy, where the sheer volume of e-commerce traffic could have overwhelmed Alibaba’s servers. But that worst-case scenario didn’t happen.
With an intelligent engine—the Daling—developed by Alibaba Group’s Institute of Data Science & Technologies, Alibaba was able to forecast the demand of data center resources hours ahead of time and automatically allocate them in a timely manner. The intelligent cloud-based system also calculated the likelihood of malfunction to maximize system usage and stability, with minimal cost.
“These technologies ensured that our systems were scaled up rapidly with expanded capacity, to as many as 100,000 servers within an hour. We were able to process 325,000 orders per second during peak traffic spikes,” Min quipped.
By the stroke of midnight at the end of Singles’ Day 2017, Alibaba had raked in a whopping US$25.3 billion, double the combined value of the US’ Black Friday and Cyber Monday sales in 2017.
Confident in its technological prowess and secure in its market dominance, especially in the e-commerce sector, Alibaba is looking to provide its AI-enabled, cloud-backed solutions to other businesses and industries. While Amazon, Microsoft, IBM and Google are recognized as incumbents in the cloud-AI-as-a-service domain, Alibaba is getting in on the action with its proprietary technology, the ET Brain.
Thinking big
At the NVIDIA AI conference held in Singapore in October 2017, Min offhandedly mentioned in his keynote address that the ‘ET’ in ET Brain stands for ‘Extreme Tech,’ not to be confused with the titular extraterrestrial in Steven Spielberg’s movie. This nomenclature is apt in more ways than one.
Embedded with visual and speech recognition capabilities that complement and reinforce machine learning, the cloud-backed ET Brain is not something alien, but rather a digital abstraction of one of the most powerful computers to have ever evolved—the human brain. While having one’s ‘head in the clouds’ once carried negative connotations of impracticality and inattentiveness, the ET Brain might just reverse this definition.
On home soil, the ET Brain has already left its mark on city management. In Hangzhou, China, where Alibaba Cloud is headquartered, the ET City Brain draws on video footage of road traffic and seamlessly combines it with data from a wide range of sources, including real-time information from the local transportation bureau and public transportation systems. It then automatically toggles the city’s traffic light network to optimize the flow of vehicular traffic, also flagging traffic rule violations to authorities.
“In [the Xiaoshan district of Hangzhou], the ET City Brain reports more than 500 traffic rule violations, accidents and congestion scenarios on a daily basis, with an accuracy of over 92 percent,” said Min.
This has enabled a 15 percent increase in traffic speed on roads in the Xiaoshan district, resulting in average time savings of three minutes per vehicle.
“Our cloud-AI program also helps identify the quickest routes for emergency vehicles, assessing their progress on roads and managing traffic lights to make their journey smoother. In the Xiaoshan district, this has halved the average time taken for ambulances to arrive at their destinations,” Min explained. What began as an endeavor to better manage a city’s traffic has inadvertently helped to save lives.
An AI on health
If Alibaba Cloud’s ET City Brain accidentally ventured into life-saving territory, its ET Medical Brain was purpose-built to explore the vast landscape of health and medicine. Consistent with the company’s bold ambitions, the ET Medical Brain is being deployed in hospitals to tackle one of the most complex diseases that has dogged humankind—cancer.
The first procedure that most cancer patients undergo is diagnostic imaging, which gives clinicians a snapshot of the offending tumor. However, sifting through the thousands of medical images can be extremely time-consuming, and depending on the experience of the pathologist, some variability in diagnostic accuracy is inevitable. These problems may soon become a thing of the past with the ET Medical Brain.
“Strategically, we are focusing on a single pain point, which is medical image analysis, a highly specialized area where demand for expertise far outstrips supply,” said Min.
For starters, Alibaba Cloud is working with hospitals to train AI software to diagnose thyroid nodules by scanning ultrasound images. According to Min, early results indicate that the AI is making an accurate diagnosis in 85 percent of cases, a 15 percent improvement over the average accuracy rate of doctors.
Meanwhile, in an impressive showcase of the potential of cloud-AI systems, Alibaba Cloud launched its Tianchi program in March 2017, a Big Data hackathon competition to crowdsource AI-assisted medical solutions. The program has since attracted 100,000 data scientists from 75 countries. The first round of the competition concluded in September 2017, yielding significant improvements in AI-assisted image analysis of lung computed tomography scans for early lung cancer detection. Some of the solutions are now being used in pilot studies at independent medical centers and public hospitals.
“In the second round of the competition, we will focus more on precision medicine,” Min said. “With genome sequencing, proteomics and metabolomics analysis becoming more and more accessible, it is possible to ‘predict’ an individual’s health status, paving the way for truly personalized medicine.”
From managing a metropolis to modernizing medicine, it would appear that the sky is the limit for AI powered by the cloud. Min thinks that it is up to us to dream up new uses for this technology, and that those who adopt it are embracing the future.
“People often use the phrase ‘seeing is believing’,” Min said. “With technologies such as our ET Brain, I much prefer the phrase ‘if you believe it, you will see it’.”
This article was first published in the January 2018 print version of Asian Scientist Magazine. Click here to subscribe to Asian Scientist Magazine in print.
———
Copyright: Asian Scientist Magazine.
Disclaimer: This article does not necessarily reflect the views of AsianScientist or its staff.
