Covering Disruptive Technology Powering Business in The Digital Age

Home > Archives > News > Alibaba Shows Off The Endless Possibilities of Data and Machine Intelligence at Apsara Conference 2019
Alibaba Shows Off The Endless Possibilities of Data and Machine Intelligence at Apsara Conference 2019
September 27, 2019 News


The second day of the Apsara Conference 2019 continued today in Cloud Town located in the beautiful city of Hangzhou, China, packed with all kinds of exciting announcements and activities. While the first day was focused on the whys related to the theme of this year’s conference, “The Rise of Data Intelligence”, day two was all about the hows – more specifically, how Alibaba Cloud and its ecosystem of partners are working hard to provide the intelligence that can feed the new and innovative demands of today and the future. There’s a lot to cover, so here goes.

The morning keynote session was hosted by Jin Ma, VP of Alibaba Cloud Intelligence, who started by giving an overview of how artificial intelligence has progressed over the years, to now becoming something that, without realising it, we can’t live without. The featured speakers that followed, all part of Alibaba Cloud Intelligence or Damo Academy arms of Alibaba Group, went in-depth into explaining about Alibaba Cloud’s own AI, or as they sometimes referred to as machine intelligence, and how it is used to invent state of the art intelligent technologies and products that will transform the world.

The AI technologies were broken down into three main vertical domains: vision, speech and NLP.

First up, Xiansheng Hua, VP of Alibaba Group, Head of AI Centre and Senior Fellow of Damo Academy, showed off how the visual technology part of AI has improved tremendously through various techniques such as better algorithms, high precision image segmentation, end-to-end quantisation, increased drill efficiency, inference acceleration in addition to various pre and post processing strategies to greatly improve the efficiency and accuracy of visual technologies for a wide range of use cases.

As Hua and many of the speakers highlighted, algorithms alone are not enough. They must be combined with industries and industry practices in order to create more value. In terms of vision AI, deep learning and highly accurate visual recognition technologies are able to make traffic predictions, as well as predict accident rates to allow local enforcers to take precautionary measures to avoid traffic accidents and save lives.

Meanwhile, 3D capabilities can now be used to build digital twins, or highly accurate representation of a city or the objects or systems that contain within it, for high precision analysis, sensing, detection and prediction.

In healthcare, AI can be used to provide better and quicker diagnostics provided from AI can be with greater accuracy, pharmaceutical and healthcare manufacturers can improve their products and patients can have a better understanding of their condition as well as manage their health better.

These are but a few examples of the possibilities of how machine intelligence could be integrated with many traditional industries, using computing power, big data and smart algorithms to create value. An interesting thought shared by Hua was that without these industries, then there would be no AI.

Machine Intelligence to Navigate Our Complex, Unpredictable World

Next, Zhijie Yan, Head of Speech Lab and Fellow of Damo Academy, Alibaba Group, shared some of the breakthroughs that Damo Academy has made over the years in the area of natural language and speech processing.

He said that voice and language AI technologies have been widely applied in the Alibaba economy, for instance through Alibaba’s chatbot or voice assistant, as well as through features such as smart translation that allows buyers to communicate with foreign merchants, or smart call desks or centres. Yan added that MIT Technology Review had ranked Damo’s technology as among the top 10 breakthrough technologies of 2019, among the likes of Google and Amazon.

Yan then demonstrated how Alibaba’s text-to-speech (TTS) solution that’s able to produce natural sounding voice on the fly. The progress made in this segment of AI means it will soon become harder for us to differentiate between human voice and that of a machine, and the days when we will be able to converse with machines using natural language, a staple of the sci-fi realm, are fast approaching.

Yan stressed on the importance of narrowing the gap between academia, R&D and industry. As such, it is common practice for the latest Damo technologies to be made instantly available on Alibaba Cloud, making it accessible and affordable for anyone; individuals, businesses and developers alike.

Realising that the barrier to entry may be too high for most, Yan commented, “Damo launched its self-learning service, so you can infuse your industry-specific data with that service. Therefore, you don’t need to know anything about language and voice AI, but you will be able to use the technology at your own convenience. This way, we’re able to put you behind the wheels and let you keep full control over your data. So in that sense, Damo’s AI is everyone’s AI.”

Besides using this technology for industry applications such as smart call centres, smart navigations, smart callouts and smart marketing conversations, he presented several other scenarios where voice and language AI features can and are perhaps already being used.

In the smart TV space, for example, users can talk to Alibaba’s AI as if they’re talking to a real human being. The system can recognise the user’s age, gender and other characteristics to provide more personalised services or recommendations. In court proceedings, Yan explained that language AI is starting to be used to replace manual transcription services, and he even shared that an AI judge assistant had already been used to conclude its first case.

After that, Gang Wang, Head of Automatic Drive Lab and Fellow of Damo Academy, Alibaba Group, talked about the journey of Alibaba’s Autonomous Driving division, and shared how it is one that is fraught with many challenges.

He attributed the most pressing difficulties are simply due to the sheer diversity and complexity of real-world traffic scenarios. Wang to tackle these problems and produce a working solution for autonomous vehicles to roam our streets in the future, the engineers and scientists at Damo Academy have had to adopt the “no free lunch” theorem.

In other words, it is not viable for them to develop a generic algorithm to work out every scenario. Instead, they have to go deep into every scenario and develop specific algorithms that account for all the scenarios possible, taking into account different motorist and pedestrian behaviours and actions, road types and conditions, weather situations, etc.

Once they are able to work out these kinks, the technology will not just bring autonomous cars closer to reality, but Alibaba’s mission is also to make logistics more convenient and efficient for various industries. Even on the grounds of the conference, several concept Cainiao autonomous vehicles roamed to offer simple services such as selling beverages much to the amusement and amazement of visitors.

Forging the Path to a Digital World Through AIoT

Of course, in order to make all this possible, Alibaba is providing the right platform or foundation, exemplified by the second-day announcement of the 3rd generation of the X-Dragon Architecture designed to advance the computing era for AI (you can read more about the launch by clicking here).

But another interesting announcement was made by Yunfei He, General Manager of Product Management, General Manager of AIoT P&S, Alibaba Cloud Intelligence, about the latest iteration of the AliOS-Things 3.0, Alibaba’s IoT version of the AliOS Family.

He believes this is an essential progress towards a future where most of the data will no longer be created by humans, but the millions upon millions of devices that we will depend on. In the near future, He said, every business will own thousands of smart devices and would therefore need a smart and reliable Artificial Intelligence of Things (AIoT) to cope, adding that Alibaba’s AIoT platform is already three years ahead of the demand curve.

With the combination of AI and IoT, He said that the user experience can be significantly optimised and the AIoT platform is able to revolutionise a host of applications involving home appliances, factories, surveillance cameras, healthcare equipment, drones, smart devices and autonomous vehicles, which are all now starting to adapt to AIoT technologies.

He then got a little philosophical when he said that the chips (such as the ones launched during the Apsara Conference), are the foundation of AI computing, or the body, and the algorithms are its soul. Data is then like the blood that flows within the body. When you have all those working together, He said, ”then your AI comes to life.”

He ended his part of the presentation with the following words. “We believe in the next 20 years, every device will enable enterprises to enter a digital world. Alibaba’s cloud and IoT capabilities will help companies accelerate their journey towards building that digital world. That is our mission.”