Tencent increases its focus on artificial intelligence

Tencent Holdings Ltd.'s new headquarters stand under construction in Shenzhen, China, on Monday, Aug. 22, 2016. The new headquarters for Tencent is a $599 million project aimed at creating a campus-like atmosphere for the urban setting. Scheduled for completion next year, the Shenzhen skyscraper could become one of the largest labs for new internet services and connected devices. Photographer: Qilai Shen/Bloomberg via Getty Images

When it comes to artificial intelligence (AI) and Chinese tech companies, thoughts often begin and end with Baidu. But Tencent, Asia’s second highest-valued tech company behind Alibaba, has reminded the world that it too is investing in the field.

Search giant Baidu was one of the first to make a major commitment to deep learning. It spent over $2.9 billion on R&D over a 2.5 year period, according to Bloomberg, and currently has more than 1,300 specialists working on a variety of technologies that include AI and augmented reality. Baidu, however, suffered a blow when its chief scientist Andrew Ng, who heads up its U.S.-based research team, announced his departure last week.

There’s one more exit to add to that list after Tencent announced today that it poached machine learning researcher Tong Zhang, who heads up Baidu’s Big Data Lab, to lead its own AI Lab. The Shenzhen-based lab is focused on computer vision, speech recognition, and natural language processing.

Tencent, best known for WeChat, China’s top messaging app, announced the lab last April. It said today that it has 50 AI specialists housed there. Aside from that development facility, Zhang — who received a PhD in Computer Science from Stanford and has worked at IBM and Yahoo — will lead a team of 200 product engineers that’s tasked with converting AI advances into tangible features and updates for Tencent’s apps and services.

“Tencent is looking at four areas for AI application: content, social, online games and cloud services. At present, over a hundred Tencent products, including Weixin/WeChat, QQ and Tian Tian Kuai Bao, a Tencent news app, use AI technology,” the company added.

That’s a similar goal to Baidu, which got serious on deep learning when it hired Ng, a co-founder of online learning startup Coursera, in 2014.

A world-renowned AI expert, Ng previously made his name when it helped found Google’s deep learning team, Google Brain. Ng said in his departure note that Baidu’s AI efforts had been felt across its “existing businesses in search, advertising, maps, take-out delivery, voice search, security, consumer finance and many more” areas.

“The team is stacked up and down with talent; I am confident AI at Baidu will continue to flourish,” he added.

For now, Tencent is talking up its AI prowess in the field of Go, the strategic game that Google made its mark on when its AI (AlphaGo) triumphed over world champion Lee Seedol last year.

Tencent said its ‘Fine Art’ AI, which was developed by 13 Tencent engineers, defeated high-ranking Japanese Go player Ryo Ichiriki last week. All in all, the firm said the AI has taken on 100 “renowned human players,” winning 406 of over 500 rounds that it has competed in.

Source: https://techcrunch.com/2017/03/27/tencent-ai/

Uber’s autonomous test cars return to the road in San Francisco today

uber-self-driving-san-francisco-sf

Uber has confirmed that its small fleet of autonomous vehicles in San Francisco will be back on the road today. The company grounded its entire self-driving test fleet in the U.S. on Saturday, following an accident involving one of its vehicles in Arizona earlier in the day.

“We are resuming our development operations in San Francisco this morning,” an Uber spokeswoman told us this morning.

The spokeswoman also confirmed Uber’s autonomous cars in Arizona and Pittsburgh remain grounded, but said the expectation is for them to shortly be back on the roads too.

Uber grounded all its self-driving test cars following the accident, initially suspending the operation in Arizona but also in the two other cities where it is currently testing the vehicles: San Francisco and Pittsburgh, pending the outcome of an investigation.

While Uber’s self-driving cars are able to drive autonomously, the test vehicles include a human driver sitting in the passenger seat so they are in a position to take over should it become necessary.

The two vehicles it’s testing in SF are apparently at a different stage of development vs the 12 test cars apiece it’s been piloting in the other locations — explaining why it’s feeling confident enough to return to the road in SF but not yet elsewhere.

Early reports of the Arizona incident appear to have cleared Uber’s technology of blame, with local police saying the accident occurred because a normal (i.e. human-driven) vehicle failed to yield to the Uber, which was in self-driving mode at the time.

The company’s self-driving test program has, however, previously faced safety-related criticism — including when one of the vehicles ran apparently ran a red light. In that instance Uber claimed the car had not been in self-driving mode at the time — though a New York Times report, citing two Uber sources, has suggested the opposite.

Beyond questions over the safety of its self-driving tech, Uber continues to face wider criticisms relating to its company culture — with accusations of sexism by a former employee continuing to pile pressure on the leadership (and apparently contributing to the decision of president Jeff Jones to leave the company earlier this month, amid the turmoil).

Source: https://techcrunch.com/2017/03/27/ubers-autonomous-cars-return-to-the-road-in-san-francisco-today/

Editas, Allergan Ink $90M CRISPR-Based Eye Drug Deal

editas

NEW YORK (GenomeWeb) – Editas Medicine has signed a deal potentially worth more than $90 million, giving Allergan the exclusive rights to license certain of its CRISPR genome editing-based treatments for eye diseases.

Under the terms of the deal, Allergan subsidiary Allergan Pharmaceuticals International Limited will have exclusive options to license up to five of Editas’ ocular drug candidates including LCA10, its preclinical treatment for Leber congenital amaurosis, an early candidate for the first disease to be treated medically with CRISPR. Allergan will also have access to CRISPR/Cpf1, an alternative DNA targeting nuclease developed by Editas Cofounder and Broad Institute Researcher Feng Zhang.

Editas will receive an upfront payment of $90 million for the development of the five drug programs, and stands to receive additional payments upon achieving undisclosed near-term milestones specifically related to LCA10. It may also receive payments tied to certain development and commercial milestones, plus royalties on a per-program basis.

Allergan will be responsible for all development and commercialization efforts for all optioned candidates, subject to an Editas’ option to co-develop and co-promote one or two of the drugs in the US.

Additional terms were not disclosed.

“The Allergan team is excited to work with colleagues at Editas Medicine to develop and potentially deliver game-changing treatment for retinal diseases like LCA10,” David Nicholson, Allergan’s chief research and development officer, said in a statement. “This program is highly complementary to our ongoing eye care development programs where unmet medical need exists for patients.”

In August 2016, Editas signed a collaboration deal with Adverum Biotechnologies to develop delivery methods of genome-editing eye therapies.

Allergan has long been a leader in advancing innovative therapies to treat eye diseases,” said Editas CEO Katrine Bosley. “Working together with Allergan through their Open Science R&D model significantly enhances our ability to develop genome editing medicines to help patients with serious eye diseases.”

Source: GenomeWeb, www.genomeweb.com

Freenome Raises $65M in Series A Funds

freenome

NEW YORK (GenomeWeb) – Palo Alto, California-based Freenome announced today that it has raised $65 million in Series A funding, which it will put toward conducting clinical trials to validate its non-invasive, early cancer detection technology.

The round was led by venture capital firm Andreessen Horowitz, which last year helped the company raise $5.6 million in seed funding. Google Ventures, Polaris Partners, Innovation Endeavors, Spectrum 28, Asset Management Ventures, Charles River Ventures, Third Kind Ventures, AME Cloud Ventures, and Allen and Company contributed to this Series A round, as did previous investors such as Data Collective and Founders Fund.

Andreessen Horowitz General Partner Vijay Pande will join Freenome’s board of directors, the company added.

“These funds will help bring Freenome’s technology to market faster by accelerating our ongoing research and clinical trials,” wrote Cofounder and CEO Gabriel Otte in a blog post.

Freenome’s early cancer detection testing utilizes genome-wide sequencing and a computational system, called Adaptive Genomics Engine, that can be trained to analyze patterns in the cell-free genetic material floating in a person’s blood.

The firm is working with 25 research partners on various studies of its technology, including the University of California, San Francisco, Moores Cancer Center at UC San Diego Health, and Massachusetts General Hospital. According to Otte, five pharmaceutical companies are also using Freenome’s system to explore personalized medicine strategies in cancer.

Once Freenome commercially launches its technology, it will enter a space occupied by Grail, which today announced it has raised more than $900 million through the first close of a Series B financing. The company is planning a second close that will bring the total funds raised in the Series B to more than $1 billion.

In January, Guardant Health — which also competes in this space — inked a multi-year deal with MD Anderson Cancer Center to make its Guardant360 test the preferred liquid biopsy test at the facility. Guardant and MD Anderson will also jointly fund interventional clinical utility studies on the non-invasive targeted sequencing approach.

Source: GenomeWeb, www.genomeweb.com