What’s happened in AI: August 6th-12th

By | August 13, 2018

Heavy week on the fundraising side. In particular, lots of Chinese AI startups raised money this week including Moviebook, Suiyuan, and Tianrang.

In addition to startup fundraising, nonprofits were big winners this week as well. AI4All, a nonprofit working to increase diversity and inclusion in artificial intelligence, received a $1mm grant from Google. It’s great to see collaboration like this, and investment from the top tech companies in increasing diversity in the industry.

Company developments:

IBM Launches EdX Certificate Programs in Deep Learning and Chatbots – Aug 9, 2018 (Campus Technology)

  • IBM is introducing two new Professional Certificate programs on the edX platform, focused on emerging tech in artificial intelligence: deep learning and chatbots. Each program offers a series of self-paced courses designed to build “critical, in-demand skills” for specific careers in the field, according to a news assignment
  • The Deep Learning program comprises five courses:
    • Deep Learning Fundamentals with Keras;
    • Deep Learning with Python and PyTorch;
    • Deep Learning with Tensorflow;
    • Using GPUs to Scale and Speed-up Deep Learning; and
    • Applied Deep Learning Capstone Project

Samsung to invest $22 billion into new growth areas like A.I. and 5G – Aug 8, 2018 (CNBC)

  • South Korea’s largest conglomerate Samsung Group is planning to invest about 25 trillion Korean won ($22 billion) over the next three years into new growth areas, led primarily by Samsung Electronics
  • Those investments would be made in four key areas: artificial intelligence (AI), fifth-generation mobile network technology, electronic components for future cars and bio-pharmaceuticals, the company said on Wednesday
  • Overall, the conglomerate said it planned to invest a total of 180 trillion won ($161 billion) over the next three years, which will include capital expenditures as well as research and development in its semiconductors and displays businesses. Most of that investment — about 130 trillion won of the total — will be spent in South Korea, the company said without giving further breakdowns

Baidu’s DuerOS AI assistant is now installed on 100 million devices – Aug 8, 2018 (VentureBeat)

  • Amazon’s Alexa. Microsoft’s Cortana. Apple’s Siri. The Google Assistant. They’re among the world’s most popular voice assistants, but a Chinese up-and-comer — Baidu’s DuerOS — is joining the ranks. Baidu yesterday announced that its conversational AI assistant has reached an install base of 100 million devices, up from 50 million just six months ago
  • DuerOS, for the uninitiated, is a suite of software developer kits (SDKs), APIs, and turnkey solutions that allow original equipment manufacturers (OEMs) to quickly integrate Baidu’s voice platform with smart speakers, refrigerators, washing machines, set-top boxes, and more. To date, its more than 200 partners have launched 110 DuerOS-powered devices
  • A few high-profile examples are Xiaoyu Zaijia (“Little Fish” in English), an Amazon Echo Show-like device with a swiveling screen and camera, and Baidu’s Xiaodu Smart Speaker, which launched in June 2018 and sold out within 90 seconds of availability

AI Chip startup Cerebras Systems picks up a former Intel top exec – Aug 8, 2018 (TechCrunch)

  • Cerebras Systems, one of the startups that has raised a significant amount of capital, is looking to continue targeting next-generation machine learning operations with the hiring of Dhiraj Mallick as its Vice President of Engineering and Business Development
  • Prior to joining Cerebras, Mallick served as the VP of architecture and CTO of Intel’s data center group. That group generated more than $5.5 billion in the second quarter this year, up from nearly $4.4 billion in the second quarter of 2017, and has generated more than $10 billion in revenue in the first half of this year. Prior to Intel, Mallick spent time at AMD and SeaMicro
  • Cerebras Systems is one of a class of startups that want to figure out what the next generation of machine hardware looks like, and most of them have raised tens of millions of dollars. It’s one of the startups that has been working on its technology for a considerable amount of time. Others include Mythic, SambaNova, Graphcore, and more than a dozen others that are all looking at different pieces of the machine learning ecosystem. But the end goal for all of them is to capture part of the machine learning process — whether that’s inference on the device or training in a server somewhere — and optimize a piece of hardware for just that

Dell EMC launches new Ready Solutions to speed up AI innovation – Aug 7, 2018 (ZD Net)

  • The new offering boasts specialised designs for machine learning with Hadoop and deep learning with Nvidia, with Dell EMC saying that offering “ready” solutions means organisations no longer need to individually source and piece together their own
  • The deep learning with Nvidia solution features Dell EMC PowerEdge R740xd and C4140 servers with four Nvidia Tesla V100 SXM2 Tensor Core GPUs; Dell EMC Isilon F800 All-Flash Scale-out NAS storage for deep learning, which enables the analysis of large datasets concurrently; and Bright Cluster Manager for Data Science in combination with the Dell EMC Data Science Provisioning Portal to set up, provision, monitor, and manage the cluster
  • While Dell EMC said machine learning with Hadoop features Dell EMC PowerEdge R640 and R740xd servers; Cloudera Data Science Workbench; Apache Spark; and Dell EMC Data Science Provisioning Engine, which the company said provides pre-configured containers allowing data scientists access to the Intel BigDL distributed deep learning library on the Spark framework

Attivo Networks releases machine learning-driven deception solution – Aug 6, 2018 (ITP.net)

  • The platform lowers the total cost of ownership by completely automating the deployment and maintenance of the most authentic and comprehensive deception environment. It accomplishes all of this transparently and without adding any agents to the production environment
  • “To successfully outmaneuver attackers, deception solutions need to be dynamic, authentic, and enticing to an adversary,” said Ray Kafity, Vice President, Middle East, Turkey & Africa at Attivo Networks. “By leveraging machine learning capabilities, Attivo Networks makes it easier than ever to deploy, manage, and operate deception by automating the creation and deployment of decoys and lures. This maintains the credibility of the environment, effectively reducing dwell time and accelerating incident response.”
  • The continuous learning and operation also determines when the solution needs to update or deploy new elements. It can also react to suspect behaviour and expand the deception surface. This prevents ‘fingerprinting’ by attackers who would then know what to avoid

China’s JD launches first overseas AI-powered store in Indonesia – Aug 6, 2018 (Petrol Plaza)

  • Located in Jakarta’s popular PIK Avenue shopping mall, the 270-square meter retail store is powered by artificial intelligence, facial recognition, and radio frequency identification
  • Using a JD.ID smartphone app, customers can just walk through an X-Mart, grab the product they need, and walk out of the store. Purchase transaction will be automatically generated through the app
  • JD’s unmanned store technology has also become increasingly sophisticated. Cameras placed throughout the space recognize customers’ movement and generate heat maps of the activity to monitor traffic flow, product selection, and customer preferences to help optimize inventory, product displays and all facets of store management

Narrative Science instantly explains Tableau dashboards as written stories to enhance analysis and communication – Aug 6, 2018 (Global Newswire)

  • Narrative Science, the leader in natural language generation (NLG) for the enterprise, today announces Narratives for Tableau™, a new extension built with the recently released Tableau Extension API and powered by Narrative Science
  • Narratives for Tableau instantly provides plain-English explanations of charts and graphs, giving analysts a quick and easy way to communicate to dashboard readers. This functionality allows users to generate interactive natural language stories that explain the insights within their data visualizations
  • Francois Ajenstat, Tableau’s Chief Product Officer, said, “This is the next logical milestone for our relationship with Narrative Science and ultimately, bringing value to our customers. In simplifying the ability for Tableau users to glean insights from their data, they are freed up to spend more time on analysis and less on administrative tasks.”

Fundraising / investment:

AI4ALL participants tell all—summer camps get girls involved in AI and tech – Aug 10, 2018 (Google Blogs)

  • With a $1 million grant from Google.org, AI4ALL can scale their nationwide summer camps that spark student interest in AI and help them build foundational technical skills. The Google.org grant will also create a new digital curriculum that will introduce students to fundamental AI concept
  • AI4ALL, a nonprofit working to increase diversity and inclusion in artificial intelligence, believes that all students should have the opportunity to learn about AI and explore its applications

Offering a white-labeled lending service in emerging markets, Mines raises $13 million – Aug 10, 2018 (TechCrunch)

  • Emerging markets credit startup Mines.io has closed a $13 million Series A round led by The Rise Fund, the global impact fund formed by private equity giant TPG, and 10 others, including Velocity Capital
  • “We’re a technology company that facilitates local institutions — banks, mobile operators, retailers — to offer credit to their customers,” Mines CEO and co-founder Ekechi Nwokah told TechCrunch
  • Mines started operations in Nigeria and counts payment processor Interswitch and mobile operator Airtel as current partners. In addition to talent acquisition, the startup plans to use the Series A to expand its credit-as-a-service products into new markets in South America and Southeast Asia “in the next few months,” according to its CEO

SenSat, a UK startup that uses visual and spatial data to ‘simulate reality’, picks up $4.5M seed – Aug 10, 2018 (TechCrunch)

  • SenSat, a U.K. startup aiming to use visual and spatial data to “simulate reality” and help computers better understand the physical world, has raised $4.5 million in seed funding — cash it will use to further develop the technology, and invest in its San Francisco office. The round was backed by Force Over Mass, Round Hill Venture Partners, and Zag (the venture arm of global creative agency BBH)
  • Launched in 2017 by founders James Dean (CEO) and Harry Atkinson (Head of Product), SenSat turns complex visual and spatial data into what is described as “real-time simulated reality” designed to enable computers to solve real world problems
  • But to do this, first the real world needs to be simulated and those simulations injected with data that computers can understand and interact with. And that starts with using new technology to photograph the real world at a level of detail that goes beyond satellite imagery

AI giant SenseTime leads $199M investment in Chinese video tech startup – Aug 8, 2018 (TechCrunch)

  • SenseTime may be best known as the world’s highest-valued AI company — having raised $620 million at a valuation of over $4.5 billion — but it is also an investor, too. The Chinese firm this week led a 1.36 billion RMB ($199 million) Series D funding round for Moviebook, a Beijing-based startup that develops technology to support online video services
  • Moviebook previously raised a 500 million RMB Series C in 2017, worth around $75 million. SB China Venture Capital (SBCVC) also took part in this new round alongside Qianhai Wutong, PAC Partners, Oriental Pearl, and Lang Sheng Investment
  • With the investment, SenseTime said it also inked a partnership with Moviebook which will see the two companies collaborate on a range of AI technologies, including augmented reality, with a view to increasing the use of AI in the entertainment industry

Self-driving truck startup Kodiak Robotics raises $40 million – Aug 7, 2018 (TechCrunch)

  • In Don Burnette and Paz Eshel’s view, trucking is the killer app for self-driving technology. It’s what led Burnette to leave the Google self-driving project and co-found Otto in early 2016, along with Anthony Levandowski, Lior Ron and Claire Delaunay. And it’s what would eventually prompt Burnette to leave Uber — the company that acquired Otto — and co-found with former venture capitalist Eshel a new driverless-trucks startup called Kodiak Robotics
  • Kodiak Robotics will use the funds to expand its team and for product development. The company has about 10 employees, according to Eshel, who was a vice president at Battery Ventures, where he led the firm’s autonomous-vehicle investment project
  • Kodiak Robotics announced Tuesday it has raised $40 million in Series A financing led by Battery Ventures. CRV, Lightspeed Venture Partners and Tusk Ventures also participated in the round. Itzik Parnafes, a general partner at Battery Ventures, will join Kodiak’s board
    • For those interested in learning more about autonomous trucks, I wrote a blog post about their impact on the logistics industry a few weeks ago

Scale raises $18 million to label data from autonomous car companies like Lyft and Embark – Aug 7, 2018 (VentureBeat)

  • Autonomous cars need data. Lots of data. According to research conducted by the Rand Corporation, they’ll have to drive hundreds of millions or even billions of miles to demonstrate their safety. Companies like Waymo are well on their way; the average self-driving vehicle is projected to generate more than 300 terabytes per year. But collecting data isn’t the only technical hurdle. Labeling it — that is, adding the markup that allows cars’ computer models to recognize and learn from it — is another
  • That’s where Scale comes in. The San Francisco-based company, which was founded in 2016 by 21-year-old MIT computer scientist Alexandr Wang, supplies an API that autonomous car manufacturers use to accelerate the data-labeling process. It today announced an $18 million funding round led by Index Ventures, Accel, and Y Combinator
  • “Our [new funding] enables Scale to rapidly advance how human intelligence and machine learning can work together to make the once arduous and manual process of creating training data a breeze,” Wang said in a statement. “The success of AI-based applications is inherently dependent on the caliber of the data inputted, and we believe our human and machine integrated system provides customers with the precision needed to power AI applications. We’re proud that leaders in autonomous vehicles (as well as other industries) have made Scale an integral and trusted part of their pioneering work.”

Tencent Leads $50M Pre-A Round In Chinese AI Chip Maker Suiyuan Technology – Aug 7, 2018 (China Money Network)

  • Tencent Holdings Ltd has led a RMB340 million (US$50 million) pre-A round in Suiyuan Technology, a Chinese artificial intelligence start-up developing cloud-based deep learning chips for AI training platforms. Existing investors including Zhen Fund, Delta Capital, Yunhe Partners and Summitview Capital also participated in the round, according to Tencent’s news portal Tencent News
  • Founded in March, Shanghai-based Suiyuan Technology develops cloud-based deep learning chips for AI training platforms. It claims that its chips will adopt self-developed architecture with high computational power, high efficiency, low cost and programmable features to support AI training and optimization. The company has two research and development centers in Beijing and Shanghai
  • The company said its main R&D team has over 15 years of premium chip and software development and mass production experience. The newly founded company has no official website or social media account yet. Information of its founders is not announced today either

Chinese AI startup Tianrang raises a $26M funding round, launches new project to apply ML to cities – Aug 7, 2018 (TechCrunch)

  • Chinese AI startup Tianrang has raised a $26 million (RMB180 million) funding round from China’s Gaorong Capital and co-lead CMB International Capital. Other investors included Ziniu Fund and Chinese fintech company Wacai. In 2016, the company raised an angel round led by Gaorong Capital and participated in by Shanghai Jindi Investment Management Ltd
  • Based on deep learning and other AI technology, Tianrang provides data analysis and smart solutions for enterprises. It was founded by in 2016 by Xu Guirong, former director of Alibaba’s Ali Cloud and chief scientist at Alibaba’s cloud platform Alimama. So no slouch on the AI front
  • Tianrang claims to be able to automatically collect and analyze marketing trends and purchase-related information on Alibaba’s e-commerce platform, allowing vendors to make better marketing decisions

MOV.AI raises $3M in seed funding to create an ‘Android for Robotics’ – Aug 6, 2018 (TechCrunch)

  • MOV.AI plans to create an ecosystem where developers, integrators and manufacturers collaborate to develop the first industry-grade O/S for autonomous intelligent collaborative robots. This could potentially produce smarter robots on a large-scale for operation and production lines. It’s now raised $3M in seed funding in a round led by Israel-based Viola Ventures and SF-based NFX
  • MOV.AI describes itself as an ‘ROS compatible operating system’. That means it enables industry-grade deployment of fleets of autonomous robots. The idea is that this will decouple the hardware from the software (a problem in robot-land), and simpler R&D, thus making robot automation affordable for any player, large or small
  • Limor Schweitzer (pictured), founder and CEO of MOV.AI says: “At MOV.AI, we have made it our mission to contribute to a world where intelligent robots perform most of the common physical tasks, which will free humankind to be more creative and productive, and enable faster market scalability. In other words we will be able to transform human operated mobile machines into autonomous robots that work safely together with people and other robots in any environment at all scalable levels”


​Optus and Curtin University partner for artificial intelligence research – August 6, 2018 (ZD net)

  • A new research group that will operate out of the School of Electrical Engineering, Computing and Mathematical Sciences has been stood up under the partnership, which will focus on the impact of AI on regional telecommunications, higher education, and the urban environment
  • The five-year arrangement will also see the appointment of an Optus chair in artificial intelligence and three Optus Research Fellows, as well as funding for PhD scholarships and student projects, Optus explained on Monday
  • Previously focusing on building cybersecurity skills, Optus Business co-invested AU$8 million alongside La Trobe University in Melbourne in late 2016 to form a cybersecurity degree that is focused on developing multi-disciplinary courses, research programs, and scholarships for students to study cybersecurity

Research / studies:

A small team of student AI coders beats Google’s machine-learning code – Aug 10, 2018 (MIT)

  • Fast.ai’s team built an algorithm that beats Google’s code, as measured using a benchmark called DAWNBench, from researchers at Stanford. This benchmark uses a common image classification task to track the speed of a deep-learning algorithm per dollar of compute power
  • Google’s researchers topped the previous rankings, in a category for training on several machines, using a custom-built collection its own chips designed specifically for machine learning. The Fast.ai team was able to produce something even faster, on roughly equivalent hardware
  • Howard’s team was able to compete with the likes of Google by doing a lot of simple things, which are detailed in a blog post. These include making sure that the images fed to its training algorithm were cropped correctly: “These are the obvious, dumb things that many researchers wouldn’t even think to do,” Howard says

MIT’s machine learning model can make cancer treatments less toxic – Aug 10, 2018 (V3)

  • Researchers at MIT have developed a machine learning technique that can lower the toxic treatments that patients are given for glioblastoma, the most aggressive form of brain cancer
  • In a simulated trial of 50 patients, the model managed to maintain the tumour-shrinking potential of treatment while lowering potency to a quarter or half of nearly all the doses. In some cases it lowered the regularity of treatments to twice a year instead of once a month
  • The researchers used a reinforcement learning technique, which is a method in which the model learns to favour behaviour that leads to a desired outcome through a system of rewards and penalties. In this case, the reward system took the form of assigning each outcome a positive or negative mark, with their size weighted on factors like chance of success. If the model chose to ‘cheat’ by simply giving patients the maximum number and potency of doses, it was marked down – forcing it to choose fewer, smaller treatments

UCI Health Opens Center for Artificial Intelligence, Deep Learning – Aug 9, 2018 (HealthIT Analytics)

  • The University of California, Irvine (UCI) and UCI Health System have launched the UCI Center for Artificial Intelligence in Diagnostic Medicine, which will leverage machine learning tools in all areas of clinical care to improve outcomes and lower costs
  • Peter D. Chang, MD and Daniel S. Chow, MD, neuroradiologists in the Department of Radiological Sciences at UCI School of Medicine, will lead the center. The team will focus on developing deep learning neural networks and applying them to diagnostics, disease prediction, and treatment planning
  • “Our goal is to empower healthcare providers, researchers and patients through the use of artificial intelligence in healthcare,” said Chang

AI spots 40,000 prominent scientists overlooked by Wikipedia – Aug 8, 2018 (The Verge)

  • AI is often criticized for its tendency to perpetuate society’s biases, but it’s equally capable of fighting them. Machine learning is currently being used to scan scientific studies and news stories to identify prominent scientists who aren’t featured on Wikipedia. Many of these scientists are female, and their omission is particularly significant in the world’s most popular encyclopedia, where 82 percent of biographies are written about men
  • The research has been carried out by an AI startup named Primer as a demonstration of the company’s expertise in natural language processing (NLP). This is a challenging but lively subfield of AI that’s all about understanding and generating digital text. Wikipedia is often used as a source to train these sorts of programs, but Primer wants to give back to the site
  • To date, the startup has identified 40,000 “missing” scientists whose coverage is similar to individuals who have Wikipedia articles, and has published 100 AI-generated summaries. It’s also been involved with three Wikipedia editathons intended to improve online representation of women in science. (Editathons are events where specialists teach one another to create and edit Wikipedia articles, usually to bolster coverage of their subject area.) And as Bohannon notes, at least one person spotted by Primer’s technology has already been given a Wikipedia article because of it — Canadian roboticist Joëlle Pineau

Researchers teach an AI how to dribble – Aug 7, 2018 (TechCrunch)

  • To do this researchers at Carnegie Mellon and DeepMotion, Inc. created a “physics-based, real-time method for controlling animated characters that can learn dribbling skills from experience.” The system, which uses “deep reinforcement learning,” can use motion capture date to learn basic movements
  • “Once the skills are learned, new motions can be simulated much faster than real-time,” said CMU professor Jessica Hodgins. Once the avatar learns a basic movement, advanced movements come more easily including dribbling between the legs and crossovers

Researchers find limitations in autonomous car technology – Aug 7, 2018 (Consumer Affairs)

  • The Insurance Institute for Highway Safety (IIHS) tested five systems in four cars — Tesla, Mercedes, BMW, and Volvo — and found variable performance in typical driving situations. Situations included approaching stopped vehicles and negotiating hills and curves. It found that the current generation of these systems are not completely reliable substitutes for human drivers
  • The researchers said they wanted to find out if these automated systems handled driving tasks the same way a human driver would. In many cases, they didn’t. Some of the discrepancies were minor, such as too-cautious braking. In some cases, however, they were dangerous. In one case, a vehicle veered sharply toward the shoulder when its sensors couldn’t detect lanes
  • “The new tests are an outgrowth of our research on Level 2 autonomy,” said Jessica Jermakian, IIHS senior research engineer. “We zeroed in on situations our staff have identified as areas of concern during test drives with Level 2 systems, then used that feedback to develop road and track scenarios to compare vehicles.”

Government / policy:

Ecuador’s All-Seeing Eye Is Made in China – Aug 9, 2018 (Foreign Policy)

  • On his second and last day in Quito nearly two years ago, President Xi Jinping, the highest-ranked Chinese official to ever head to Ecuador, made a largely overlooked visit to a boxy government facility. Xi was driven up a small hill to the headquarters of one of Ecuador’s proudest public safety achievements: a national emergency response and video surveillance system built entirely by Chinese companies and financed by Chinese state loans
  • Since Xi’s tour in late 2016, ECU 911 has expanded significantly in scope and sophistication. Initially funded by a $240 million Chinese loan in 2012, the system has a nationwide network of 4,300 surveillance cameras, 16 regional response centers, and over 3,000 government employees diligently watching video footage and responding to millions of 911 calls every year. With the help of Chinese technology, ECU 911 now boasts thermal cameras that monitor snowcapped volcanoes for signs of activity, drones capable of night vision, an automated platform for sending video evidence to courts, and an artificial intelligence research lab inaugurated by Xi himself. It is also testing large-scale uses of facial recognition to catch suspects in major cities and their airports
  • Credited with cutting crime and saving lives after natural disasters in a country long troubled by such issues, ECU 911 is portrayed by China as a showcase for its technological prowess and humanitarian impulses. And Ecuadorian politicians eager to show off new, hi-tech infrastructure to voters have been grateful. “The help from China is immense, and we only have words of gratitude, and it’s a great honor to welcome the president of this great country,” said then-Ecuadorian President Rafael Correa during Xi’s visit in 2016


New genre of artificial intelligence programs take computer hacking to another level – Aug 8, 2018 (Reuters)

  • State-of-the-art defenses generally rely on examining what the attack software is doing, rather than the more commonplace technique of analyzing software code for danger signs. But the new genre of AI-driven programs can be trained to stay dormant until they reach a very specific target, making them exceptionally hard to stop
  • That warning from security researchers is driven home by a team from IBM Corp. (IBM.N) who have used the artificial intelligence technique known as machine learning to build hacking programs that could slip past top-tier defensive measures. The group will unveil details of its experiment at the Black Hat security conference in Las Vegas on Wednesday
  • We have a lot of reason to believe this is the next big thing,” said lead IBM researcher Marc Ph. Stoecklin. “This may have happened already, and we will see it two or three years from now.”