At the same time, governments of all countries are also racing to launch advanced artificial intelligence research, which views artificial intelligence as a means to achieve greater economic impact and become an economic power.
We are in the early stages of a drastic shift in the labor market and speculation about machine learning may start to fade - because machine learning has in fact infiltrated every major area of the software industry, from simple calendars to search engines to sales Management software can see the shadow of machine learning.
Artificial intelligence is now even able to beat the world champions.Humanoid also completes the perfect backflip and lands successfully.However, artificial intelligence algorithms are far from perfect on some basic tasks that are easy for humans, such as understanding images In the scene, or identify the context of the dialogue.
At the same time, the promise of generic artificial intelligence - or the artificial intelligence of unsupervised rapid learning new tasks - remains unknown.Although a handful of companies, including Vicarious, System and Kindred, are still investing in generic artificial intelligence, But there is almost no detail or real evidence.
Today's AI applications focus only on a very narrow range of tasks, but it is these artificial intelligence-driven, narrow-minded tasks that reshape commerce, markets and industries.
To help people grasp the status quo and development trajectories of thousands of AI companies around the world, CB Insights recently released its AI report predicting 13 kinds of artificial intelligence trends that will emerge in 2018, with a view to becoming a leader in the field of artificial intelligence Bring inspiration.
First, the emergence of a new blue-collar work - robot nanny
In the United States, employment opportunities in industrial robots and manufacturing are on the rise.
Manufacturing jobs have been criticized for their often outsourced labor to cheap developing countries, and to reduce the cost of industrial robots there is also a time when production sites need to be brought closer to where they are needed.
Recently, Chinese T-shirt maker Tianyuan Garment Company signed a Memorandum of Understanding (MoU) with the U.S. government of Arkansas to set up a new garment factory in Arkansas to hire about 400 workers at an hourly wage of $ 14. Action plan to be carried out by the end of 2017.
The new Tianyuan plant in Little Rock, Arkansas, will use a sewing robot from SoftWare Automation, a Georgia-based start-up for artificial intelligence, developed by SoftWare Automation for the production of clothing by Adidas. Many seemingly strenuous work will be done by the robot , While human workers take over high-end jobs that include robot maintenance and operations, which means the number and nature of manufacturing jobs will be very different from 2008.
The Bureau of Labor Statistics defines and analyzes the different jobs in the manufacturing industry, for example, because of the impact of automation, the Bureau is not optimistic about the prospects of such positions as quality control inspectors, assemblers and builders.
In 2012, the contract between Advanced Planning and Research Agency of the U.S. Defense Department and SoftWear Automation made it clear that "the ultimate goal is to achieve a complete production facility where the production of apparel is directly man-made zero."
But consumers' changing preferences and the inability to adapt to the dramatic process changes are still completely automated obstacles.
This is evident even in the highly automated Amazon warehouse.
Amazon's collaborative warehouse robots take on the most heavy work, while human workers focus on meticulous work such as picking goods from shelves and inserting them into separate orders.
However, the robots are still far from satisfactory in terms of picking, picking and handling items in an unstructured environment.Amazon has used more than 100,000 robots in a variety of warehouses, but at the same time has created a number for humans in new distribution centers Thousands of new jobs.
Second, the application of artificial intelligence into all walks of life
The development trend of artificial intelligence is irresistible. From brewing beer to cannabis industry, machine learning seems omnipotent.
Artificial Intelligence is everywhere, and specifically speaking, machine learning is everywhere.Machine learning refers to training algorithms on large-scale data sets to let machines learn how to recognize and generate the patterns they need.As time goes by, algorithms- The right parameters are provided by human creators - they perform better in their tasks.
As long as there is data that can train the software and have the desired output in mind, this technique can basically be applied to everything.
Therefore, you will see:
British company IntelligentX wants to launch the world's first artificial intelligence beer brewing.
Russia's DeepFish uses neural networks to identify fish species and combines radar technology with artificial intelligence to distinguish fish and noise in radar images.
Hoofstep in Sweden raised funds for venture capital and realized a horse behavior analysis based on deep learning.
Are You Vegan, Gluten-Free, or Are You Allergic to Soy? New York-based Prose wants to use artificial intelligence in custom hair care products from well-known winds such as Forerunner Ventures, Lerer Hippeau Ventures and Maveron The company raised $ 7.57 million in funding.
In addition, artificial intelligence is being used in cannabis technology, DeepGreen uses computer vision to identify the genitals and health status of cannabis plants, and Weedguide raised $ 1.7 million in funding recommendations for Artificial Intelligence personalized cannabis.
The difference between hobbies and income-generating ideas is only whether one thing is viewed in the long run, and we expect to see more "AI for X" out of the box in 2018. The growing novelty of popularity and ideas shows that artificial intelligence is not a rarity and instead is one of the cornerstones of modern software and applications.
Third, China and the United States compete for global artificial intelligence leaders
Although China AIEC has a 9% share of the global market, in 2017, nearly 50% of the global IPO funds went to China for the first time beyond the United States.
China is actively implementing a well-designed vision of artificial intelligence in which China has already defeated the United States in some areas of artificial intelligence.
The Chinese government is also promoting a future AI plan covering everything from smart agriculture, smart logistics to military applications and artificial intelligence to job creation.
Some of the resources will be used in innovative Chinese startups developing artificial intelligence in various industries, ranging from healthcare to the media.
In fact, China accounted for only 9% of the global artificial intelligence startups turnover, but in 2017, the global artificial intelligence start-up financing flows to China accounted for 48% of the total investment funds, for the first time more than the United States dollar share of the United States to know that in 2016 China Only 11.3% of global funds.
In terms of the share of assets in AI start-ups, the United States still holds the dominant position globally, but its share of global transactions is declining.
In addition, the patent applications of Chinese enterprises also reflect the research and development capabilities.
In the case of patent applications, Chinese companies have largely surpassed the US The search for artificial intelligence-related patents in China far exceeds the number of patents published by the U.S. Patent and Trademark Office, based on the keyword search of titles and abstracts.
Taking deep learning as an example, China publishes 6 times more patents in this area than the United States. (Note: Before the patent application was announced, the patent filing process took a long time to come.)
Face recognition and artificial intelligence chips are also two technologies that have helped to push the development of artificial intelligence in China, the former catering to the government's plan to implement surveillance nationwide, which poses a direct challenge to the chips made in the United States.
China's unicorn company Face ++, Shangtang Technology and start-up cloud from the science and technology is the three major players in this area (the latter won the Guangzhou municipal government funding of 301 million U.S. dollars).
Nearly 50 cities in China have joined the "Shining Project" in 2017. Monitoring cameras installed in public areas and in private areas will be centrally handled for monitoring personnel as well as various situations, and media reports said the move will help China Social credit system to consider citizens' "credit".
Kuang as technology has won the support of China's insurance companies (Sunshine Insurance Group), government organizations (Russia-China Investment Group) and the corporate giants (Foxconn, Ant Financial), the company has received facial data of 1.3 billion Chinese citizens.
Alibaba, which operates through ant clothes, and Foxconn, both investors, partnered with Hangzhou in 2016 to launch the 'City of the Brain' project, which uses artificial intelligence to analyze data on surveillance cameras and social media.
Ant gold clothes in Alibaba-owned retail stores using facial recognition technology independent payment.
The United States and China are also vying for the dominance of artificial intelligence chip technology.
In June 2017, the Chinese government said that by 2020 artificial intelligence will catch up with the United States and exceed the United States as a world leader in artificial intelligence by 2030. A government-backed project aims to create a world where operations and energy efficiency are far behind Ultra-Nvidia GPU20 times the chip.Chinese company Cambrian promised to develop the chip with one billion processing units in the next three years, the company is developing a chip dedicated to deep learning.
Chinese tech giants such as Baidu and Jingdong are also investing in overseas AI companies, including the United States.
Recently, Baidu and Jingdong invested in Zest Finance, a U.S. financial technology company, and Tencent invested in Oben, a New York-based artificial intelligence firm that starts a business in China and the United States with start-ups such as Biotech and Pony.ai to further narrow the gap between the two countries The competition gap.
Although Chinese companies are aggressively looking for cooperation or investment in the United States, AIs in the United States have more Chinese investment compared to relatively fewer AIs in China.
Fourth, rely on the future of artificial intelligence defense
The Artificial Intelligence Network Security market is getting hotter, with some start-ups even having a roster of government clients hoping to take the lead over hackers.
The data center is becoming a new battlefield.
In 2014, Amazon built a customer cloud computing service for CIA to meet the stringent compliance and regulatory requirements for sensitive data.
In the fourth quarter of 2017, AWS will open these tools to government customers outside the intelligence service.
Amazon also acquired two AI network security firms, Harvest.ai and Sqrrl, to secure cloud-sensitive data.
Regardless of whether Amazon or any other start-up deliberately caters to government clients, artificial intelligence is emerging as the backbone of government-backed cyber security.
During the Cold War, the government was discussing the "missile gap" with competitors and the merits of nuclear warheads. Nowadays, the government is increasingly concerned about their differences in network capabilities, resulting in continued cyber-security and traditional national defense Fusion.
The dangers of data breaches are striking: from the disclosure of millions of social security codes at Equifax, a U.S. credit-rating agency, to events such as WannaCry ransomware and Russia's intervention in the U.S. presidential election.
An analysis of SecurityScorecard, a New York-based company invested by Intel Capital and Moody's and other companies in 2017, reported that the U.S. government organization received the lowest score on cybersecurity, with a total of 552 Local, state and federal agencies, each with over 100 public-facing IP addresses. "
Network security provides a real application opportunity for artificial intelligence algorithms, because cyber attacks will evolve continuously and protection is constantly facing a variety of unheard-of malware.Artificial intelligence can be widely used in millions of events Screening identifies abnormal, dangerous, and potentially threatening signals.
Currently, the market has a large number of emerging cybersecurity companies, are trying to push machine learning to the next level.
In the past five years, 134 start-ups have raised $ 3.65 billion in funding, and for the first time last year some 34 companies were the first to receive financing, with more companies in the market today such as Cybereason, CrowdStrike, Cylance and Tanium, each of which The market capitalization of more than 900 million US dollars.
Even traditional consulting firms such as Accenture are constantly developing artificial intelligence cyber-security technologies to better serve their federal government customers Endgame, a startup that owns customers such as the U.S. Air Force, sold its government services business to Accenture, Aroused widespread concern.
In-Q-Tel, the investment arm of the intelligence agency, invested in Anomali, Interset and Cylance in 2016. British company Darktrace claims that it has deployed 3,000 systems worldwide, including government agencies, and Logrohythm in Colorado is also working with the U.S. Air Force , NASA and defense contractor Raytheon.
Other top defense contractors are also investing constantly.
Lockheed Martin is an early Cybereason (currently $ 900 million) investor, and in 2017, Boeing invested in SparkCognition, a Texas-based cybersecurity company, through its investment arm Horizon X.
How are you, Alexa?
Amazon Echo and Google Home dominates the smart home speaker market, but the giants are less than enthusiastic about non-English speaking markets.
Alexa opened up a revolution in speech.
Voice-enabled computing became a boon at 2018CES, and IoT devices that did not connect to Amazon Alexa or Google Home were virtually non-existent.
Samsung is developing its own voice assistant, Bixby, and hopes to have all the company's products networked and intelligently powered by Bixby by 2020. LG has all of its applications connected via WiFi in 2017. Currently, more than 80 LG's product to achieve the docking of Google Home.
Although Amazon was initially a leader in voice computing, it is one step behind its language support.
Last quarter, Amazon announced that it will be shipping Alexa powered speakers in about 80 countries, but the downside is that it wants users across the globe to interact with the speakers in English, German or Japanese.
Google Home supports English, German, French and Japanese Apple's HomePod currently only supports English, but it is planned to support German and French soon.
In this regard, Google has a greater advantage than Amazon.A Google phone helper on Android phones supports English, French, German, Italian, Korean, Spanish and Portuguese.It's speech recognition capabilities - for voice - text Conversion and voice search, can support 119 languages.
Currently, the Spanish smart home market has not been given enough attention by tech giants, though it is one of the most widely used languages after the Chinese language.
In China, Alibaba said the Chinese voice speaker Tmall Genie sold more than 1 million units since its launch in June 2017.
In 2018, voice assistants will continue to compete in the non-English voice market for market dominance.
Six, white-collar workers in the process of accelerating automation
The white-collar workers here include lawyers, consultants, financial analysts, journalists, traders and others. The impact of artificial intelligence on these people is as big as that of the blue-collar workers.
More and more AI-bizarre automation and enhancement software are bringing people to a new era of artificial intelligence-assisted manufacturing or AI-optimized production, and these AI-capable tools for optimizing production are threatening the desk of white-collar work.
The chart below shows the EAAS market, where you can see that AI EAAS start-ups are in all walks of life, and in particular, lawyers, journalists, health managers , Traders or consulting industry practitioners, there is a corresponding AI EAAS software can be used.
For example, in legal work, AI has great potential to save time and improve efficiency, and natural language processing and text analysis techniques can sum up thousands of pages of legal documents in minutes, This was done in a matter of days before it took one person to work, and the use of artificial intelligence helped to improve the accuracy of your work.
As the AI platform is becoming more and more efficient and commercialized, charging patterns for third-party law firms that used to charge hourly rates will also be affected.
Programmers are not immune to many early AI projects that are focused on AI-based software testing, Debug, and basic front-end development. DiffBlue, which is based in the United Kingdom, received a big bump in finance last year and the company's business is Use AI technology for bug fixes in everyday coding efforts, client-side code writing, translating code written in one programming language to another.
The health and education industry is considered the industry that has the least impact from AI because both industries have a large number of dynamic tasks and practitioners in both industries often require a higher emotional intelligence. However, artificial intelligence is still in the midst of these two Industry penetration, education, for example, start-up companies are working to provide artificial intelligence support services, such as marking, language teaching, composition and so on.
Seven, artificial intelligence migration to the end
The artificial intelligence industry showed a clear trend to migrate to terminals in 2017. For example, AI is embedded in smaller devices and sensors and runs on the edge of the computing network. In other words, AI Leave the cloud, or even leave the phone, in turn, exist in your headset.
Artificial intelligence is increasingly dispersed.
Device intelligence, such as intelligence on smartphones, cars, and even wireless devices, enables faster, localized, scene-based information processing because there is no need to communicate with the cloud or server.
For example, a self-driving car needs to react in real time to the conditions of the road, the decision-making process is very time sensitive, and signal delays can be life-threatening, or, in the case of a private artificial intelligence assistant trained on a local device, those helpers will be able to recognize your unique Accent and your personal facial features.
In 2017, terminal intelligence has made a qualitative leap thanks to the strong investment by technology giants.
Apple released the A11 chip with the neural engine, which will be used on the iPhone 8 and iPhone X. Apple said the chip can run machine learning tasks at speeds up to 600B per second and drives machine-learning tasks like FaceID Etc. During the use of the FaceID feature, the mobile phone does not need to upload and store any user data to the cloud by emitting invisible light to the user's face.
Intel, the mainstream processor maker for most data centers, has had to catch up with the trend of terminal intelligence through acquisitions, and Intel recently introduced device-side vision computing chip Myriad X, a chip originally acquired by Intel in 2016 The chip developed for Movidius' company.
Intel said Myriad X is capable of deep learning tasks running on a variety of end devices ranging from smartphones to child monitors to drones.
Google has come up with a concept similar to its Federated Learning, except that some of its machine learning tasks run on end devices and is currently being tested on the Google keyboard Gboard.
Although terminal artificial intelligence weakens the problem of information delay, terminal intelligence has limitations in storage space and computational power compared with the cloud.
In addition, more hybrid deep learning models will emerge to allow better collaboration between different endpoints and between endpoints and the cloud.
Eight, the rise of the capsule network
Deep learning is the driving factor behind most current AI applications, and thanks to the capsule network, deep learning is now overhauled.
Different neural networks have different structures. The most famous network structure in today's deep learning is the Convolutional Neural Network (CNN). Now a new network structure, the capsule network, has become popular and has many aspects Super CNN ability.
Despite the success of CNN in recent years, we still can not ignore its shortcomings, and in many cases, CNN is underperforming and there are potential security breaches. Researchers have long been trying hard to upgrade artificial intelligence algorithms in an attempt to overcome these problems .
Let's take one of the most common examples. In face recognition, CNN learns about all the elements of the human face (eyes, nose and mouth), but can not remember the specific location of each element, resulting in the following two Figure can be considered a human face.
Geoffrey Hinton, one of the leading researchers in deep learning, released a research paper in 2017 that introduced the concept of a capsule network, or CapsNet.
This essay is still in the assessment stage and lacks sufficient tests in actual settings, but its powerful capabilities have caused quite a stir in the media and science and technology circles.
As we will not go into details here, in short, the capsule network recognizes things from higher-dimensional features, requires less training data and is less error-prone, as in the example above, with the mouth long above the eyebrows The face will be easily identified, but CNN has no way to do that.
Another problem with the CNN is that it can not handle many variations of the input data: for example, you need to take many photos from different angles of the same object as input data to train a convolution neural network to recognize the object Identifying a wide variety of objects requires a huge amount of training data.
At this point, the capsule network is said to perform better than CNN. The capsule network requires less training data and can infer additional states from several states of an object without the need to enter every state data.
Hinton also mentioned in his paper that the capsule network has undergone some complex confrontational attack tests (with some unacceptable photo-fooling algorithms) and concluded that performance exceeds convolutional neural networks.
With some simple handling, hackers can fool the convolutional neural network, and researchers from Google and OpenAI have demonstrated this with examples.
One of the most famous examples is the fact that in a 2015 essay, researchers treated an invisible glimpse of a photo of a giant panda and identified it as gibbon with a 99.3% confidence rating.
Nine, six-figure salary of artificial intelligence talent war
In short supply, the number of top-notch researchers in the field can reach millions.
China is recruiting experts in the field of artificial intelligence.
Some of the top machine learning researchers listed in BMW China are earning about $ 56.7 to $ 624,000 and other companies are giving $ 31.5 to $ 410,000 to machine learning specialists. The job offers are based on recruitment platforms in China Hire website.
According to a recent Tencent report, there are an estimated 300,000 people currently in the field of artificial intelligence, including students in related fields of study, while companies may need one million or more people Intelligent experts to meet their engineering needs.
In the United States, searching for Artificial Intelligence on Glassdoor, the workplace community, shows more than 32,000 jobs, many of whom earn six figures.
Large companies in order to dig the best artificial intelligence talent, will naturally give the most competitive salary.
DeepMind, which was acquired by Google in 2014, said in its financial report that "staff costs and other related costs" were 104.8 million pounds last year.A quick search of LinkedIn employees had 415 employees and it was assumed that this was the 2016 team's Scale, after deducting other expenses, the average salary of a team member is £ 252,000 (about $ 350,000).
In addition, artificial intelligence researchers at major technology companies have also left and started to set up their own companies.
Ng set up an artificial intelligence fund and raised $ 175 million after he left Baidu, and the chief technology officer of Groq, an artificial intelligence chip startup, developed TPU at Google Hardware Engineering and later at Google X.
Yu Kai, chief technology officer and co-founder of Horizon Robot, a domestic start-up company, also worked for Baidu as head of Baidu's Deep Learning Research Institute, leading the image recognition team.
Undoubtedly, the battle for talent will become more intense as the talented people continually flow to start-ups.
Ten, speculation in machine learning will subside
Machine learning will soon "get out of the altar." More than 1,100 new AI startups that have emerged since 2016 need a solid business model to stay alive.
Big data first, then cloud, is machine learning now, and the tech boom comes in waves.
In 2017, the popularity of machine learning ushered in the peak.
This year, incubators spawned more than 300 AI start-ups, triple the number of 2016. In that year, investors invested more than $ 1.52 billion in artificial intelligence startups in various fields, raising funds in 2016 141%.
Since more than 1,100 emerging AI companies have completed their first round of financing since 2016, in this perspective, it is more than half of all AI start-ups that historically had equity financing. From this perspective, there are More than half of the historic AI start-ups completed the financing.
However, this wave of speculation soon subsided.
The normalization of machine learning will make investors picky about the AI companies they fund.
As Frank Chen, a well-known venture capitalist, a16z said, "There will be no investor looking for an AI startup in a few years." Startups use the necessary artificial intelligence algorithms to power their products as a "hypothetical."
In fact, we have seen this in many industries.
Machine learning is inseparable from IIoT. We need artificial intelligence to understand the massive amounts of data collected in machines and sensors and process them in real-time. Almost all cybersecurity companies are using machine learning techniques to some extent. In addition, large technology companies provide organizations with a machine learning solution.
Top investors are carefully evaluating startups using artificial intelligence technology, for example, freenome, the liquid biopsy diagnostics company, received five unlabeled blood samples before getting the a16z investment intent and started an analysis using artificial intelligence algorithms.
Eleven, Amazon, Google, Microsoft dominates Enterprise AI
Within five years, investors have invested 180 million U.S. dollars in startups focused on enterprise AI services. Now, Amazon, Google and Microsoft may face small companies being eliminated.
As more and more companies are dedicated to integrating machine learning into their products, start-ups are also starting to offer ML-as-a-service.
Currently, large tech companies such as Google, Amazon, Microsoft and Salesforce are struggling to improve their enterprise AI products and space smaller companies and funds.
Google introduced Cloud AutoML, which lets users train their algorithms with their own data to meet specific needs.
Amazon is launching the AI-as-a-service with an Amazon AI slogan under the AWS banner. The goal of Amazon AI is to serve large or small developers who need artificial intelligence without having to prepay or get involved Much more trouble Amazon launches a product-like API that allows developers to access Amazon Lex (Amazon's NLP capabilities), Amazon Polly (Amazon's speech synthesis capabilities), and Amazon Rekognition (Amazon's image analytics capabilities)
In the fourth quarter of 2017, Amazon extended its reach to include video recognition, audio transcription, and sentiment analysis, and left a deep footprint on the evolution of AWS, with revenues of only $ 5 billion in the fourth quarter, up 44% .
In addition, Microsoft and Amazon are also fierce competition, followed by Salesforces and Oracle and other companies.
Twelve, artificial intelligence diagnosis received the approval of regulatory agencies
Machine learning will soon become a routine operation in the field of medical imaging and diagnostics.
U.S. regulators are considering approving artificial intelligence for clinical use.
The value of artificial intelligence in diagnosis is mainly reflected in the early detection of illness and improve the accuracy of the area.
Machine learning algorithms can compare the medical images of millions of other patients for negligible nuances to the human eye, and the algorithm can do it in seconds, but humans can take hours .
There are also a handful of AI-monitoring tools for consumers, such as SkinVision, which uses computer vision to detect suspicious skin diseases and a new wave of artificial intelligence medical applications to hospitals and clinics.
Recently, the global biopharmaceutical company AstraZeneca announced its cooperation with Alibaba's Ali Health to develop artificial intelligence-assisted diagnostics and screening applications in China.
Prior to this, General Electric and NVIDIA tried to bring deep learning technology to the medical field, and Google DeepMind also tried to use artificial intelligence technology to detect eye diseases.
The entry of giants such as Google's DeepMind, IBM, General Electric Corp. and Alibaba have made it more difficult for start-ups to divide up the market's cake, but this has not stopped the start-ups from taking risks.
Healthcare is still one of the hottest areas of artificial intelligence venture capital, and the continued growth of many companies focusing on medical imaging and diagnostics has contributed to this result.
Arterys, the maker of medical imaging startups, received the first FDA approval and reportedly approved its cloud computing platform for analyzing heart images after a series of tests on accuracy and diagnostic speed. Arterys is currently applying for FDA approval of AI Application in oncology.
Another Israeli startup called MedyMatch uses in-depth learning techniques to analyze intracranial hemorrhage to analyze CT scans, and recently the FDA was given groundbreaking qualification to accelerate its time-to-market.
Among the most controversial areas for high-risk industries such as healthcare are those who assume the responsibility for misdiagnosis of artificial intelligence systems, which are currently aiding radiologists and physicians, and will not be the ultimate adjudicators of diagnostics .
Thirteen, artificial intelligence to DIY
Let your voice assistant sound like a movie and television drama or create your own AI camera.
You do not need to have a Ph.D. in computer science or math and you can build your own AI system.
At present, a large number of open source software, massive APIs and SDKs on the market and Amazon or Google kits that are easy to get started drastically reduce the barriers to people entering the field of artificial intelligence.
Google has introduced AIY (artificial intelligence yourself) plan, designed to allow users of all ages to DIY their own artificial intelligence products.
The AIY Voice Kit, the first product born on the AIY project, is a speech recognition kit that incorporates the Raspberry Pi. To help voice assistants, like the characters in the BBC science-fiction drama "Doctor Who," use the 80's Communication and intelligent assistant interaction. (In the play, call themselves "Dr." Lord of the lords with his disguised as the 1950s British pavilion time machine Tadis and his partner in time and space to explore leisurely, punish evil , Save civilization, help weakness.) Not hard to find, users are based on artificial intelligence technology to create more new inventions.
In addition, Google also introduced AIY Vision Kit, which supports neural network model, you can use algorithms to identify cats and dogs, but also to match the facial expressions and moods.
Amazon also introduced DeepLense, a $ 249 deep learning camera that Amazon provided for $ 7,500 for the winner of the first DeepLense hacker marathon.