Five AI China Startups Compete for Glory, and a DGX StationSeptember 27, 2017
Five CEOs of startups took the stage yesterday at GTC China in Beijing to see who could be the next Jack Ma.
They were ostensibly competing for an NVIDIA DGX Station. But the real prize was to be recognized as one of China’s most promising AI startups. The CEOs described their company and business model before a live audience and a panel of four judges, including prominent China VCs and NVIDIA execs. Then, the judges took a vote and a winner was declared.
The companies were focused on a diverse range of challenges, but they all agreed on four things:
- AI deep learning offers the best path to solve hard challenges
- China’s market opportunity is enormous
- China offers a rich pool of developer talent
- Their solution aims to assist, not replace, trained professionals
Participating companies were:
- Naturali – A conversational interface for mobile devices that makes it easy for users to perform tasks with speech commands.
- 12Sigma – Improved accuracy of patient diagnoses for cancer, heart disease and lung disease (clinically known as chronic obstructive pulmonary disease).
- Imperial Vision – Smart compression to reduce storage costs and improve efficiency of image and video delivery.
- Dr. Brain – Personal diagnosis of brain disorders and diseases like Alzheimer’s.
- Kavout – Stock investment analysis and recommendations to assist investment advisors.
All five are focused on solving a difficult challenge. Dr. Brain’s CEO, Lawrence Wang, said China has the largest population of aging people: approximately 240 million individuals. By 2030, it could cost China as much as $1 trillion to diagnose and treat those with Alzheimer’s. His company is working with nearly two dozen hospitals in China to build a database in the cloud to make it easy for trained medical staff to quickly diagnose the disease.
Alex Lu of Kavout said asset managers and buy-side analyst firms are challenged daily with three questions: when to buy a stock, when to sell and how much to invest? Its algorithms combine technical analysis with investor sentiment to help clients manage $10 billion in assets under their management.
And the Award Goes to …
After a brief vote by the panel of judges, NVIDIA Vice President of Business Development Jeff Herbst announced that 12Sigma founder Xin Zhong had won the day’s prize.
Zhong, who had formerly worked at Goldman Sachs, Merrill Lynch, Morgan Stanley and Qualcomm, was focused on one of the toughest challenges: fighting lung cancer, the leading cause of cancer death in both men and women. Each year, more people die of the disease than of colon, breast and prostate cancers combined.
Early detection and diagnosis are crucial to reduce mortality. The challenge is that CT scans add 10 times the image workload for radiologists. 12Sigma uses GPU deep learning and computer vision to automatically detect pulmonary nodules from CT scans. The company is working with just shy of 40 hospitals in China to gather patient data to cut radiologist’s time spent evaluating imagery in half. The company’s approach has achieved diagnosis sensitivity of 96.5 percent, with a lower level of false positives than radiologists.
12Sigma’s offices in San Diego and Beijing are loaded with desktop PCs packed with three-way and four-way NVIDIA TITAN GPUs. Zhong said “without GPUs we can’t do anything – they are 50-60x faster than CPU-based systems.” Hopefully the company’s shiny new DGX Station will accelerate its life-saving work.