Artificial Intelligence Panorama and Development Trend Analysis in 2018

The panorama covers the infrastructure of the big data artificial intelligence industry, open source frameworks, data APIs, data resources, cross-infrastructure analysis, industrial applications, enterprise applications, analytical tools, etc., covering 1095 big data companies into the panorama.

Some key companies appearing in the panorama are on the market, especially Cloudera, MongoDB Pivotal and Zuora. At the time of this writing, others are preparing to go public, such as Elastic.

2018 artificial intelligence big data development trend

2018 is an exciting but complex year in the data world. On the one hand, data technology (big data, data science, machine learning, artificial intelligence) continues to grow and become more efficient, and is widely used by companies all over the world. So far, one of the key themes of the business community in 2018 is “digital transformation,” which is no accident. This word may be a bit strange for some people, they will be embarrassed: Is this not something that has been going on for the past 25 years? But it just reflects the fact that many traditional industries and companies are now fully committed to a true data-driven journey. On the other hand, the broader public community is already aware of the flaws in data. Whether through public debate on artificial intelligence risks, Cambridge Analytica scandals, large-scale Equifax data breaches, privacy discussions related to gdp, or reports of increasing government surveillance activities in China, the data world has begun to be exposed. Some darker, more terrible hidden dangers.

1) Infrastructure and analytical tools

From an industry perspective, the data ecosystem is still as exciting and vibrant as ever, with a wealth of innovative start-ups, proven “scale expansion” and many active public technology providers. Most importantly, many large and small customers are applying these technologies on a large scale and gain undeniable value from their efforts.

As the cycle of replacing older IT technologies with more modern data products continues, the big data market (infrastructure, analytics) seems to be rapidly cycling through most of the early buyers and gradually transitioning to the late stages of traditional adoption curves.

In addition, the data world continues to move in the direction of the cloud. Considering the growth rate of large public cloud service providers, it is shocking to generate billions of dollars in revenue every quarter. This trend has led to continued focus on supplier lock-in, which may provide opportunities for start-up companies that offer cloudy solutions. However, so far, companies that adopt a cloudy strategy still tend to rely on one supplier as their primary provider.

As their business continues to evolve, large cloud providers offer a wide range of big data, data engineering, and machine learning tools through their platforms (such as Amazon Neptune, Google AutoML, etc.), often with aggressive pricing strategies and mutual The competition is getting more and more intense, all in order to attract more developers, because their real business model is data storage. As the scope and maturity of such tools continue to increase, this has had a major impact on the data technology arena. It can be said that start-ups are more difficult to compete with, at least in the face of broad, horizontal opportunities. Product announcement lists (such as AWS re:Invent), which are released annually at large cloud vendor conferences, can create a huge shock for start-ups as they compete directly with dozens of VC-backed startups. It will be interesting to see how the public market responds to the upcoming Elastic (an open source software company) IPO.

However, as long as start-ups are sufficiently differentiated, they still have many opportunities. In this space, many companies are rapidly expanding, with many particularly interesting and fast-growing parts of the ecosystem's infrastructure and analysis, including flow/real-time, data governance, and data structure/virtualization. The surge in interest in artificial intelligence has also led to tremendous opportunities in artificial intelligence chips, GPU databases, artificial intelligence DevOps tools, and platforms that can deploy data science and machine learning in the enterprise, as well as substantial funding.

2) Machine learning and artificial intelligence

In the field of artificial intelligence research, this is undoubtedly a crazy year, from the power of AlphaZero to the amazing speed of the release of new technologies – generating new forms of confrontation networks, alternative recurrent neural networks, and Geoff Hinton's new capsule network. Artificial intelligence conferences like NIPS have attracted 8,000 people, and thousands of academic papers are submitted every day.

At the same time, the pursuit of AGI is still elusive, which may be a thank-you thing. Most of the current excitement and fear of artificial intelligence stems from the impressive deep learning performance since 2012, but in the field of artificial intelligence research, there is an emotion that is increasingly pervasive among people: “What to do next "Because some people question the basis of deep learning (backpropagation), while others want to be able to transcend what they consider to be "brute force" methods (large amounts of data, large amounts of computational power), perhaps more inclined to adopt more neural-based scientific methods.

In the field of artificial intelligence research, many people are not worried that robots dominate the world. Instead, they worry that the continued excessive speculation in this field may eventually disappoint and lead to the arrival of another artificial intelligence nuclear winter.

However, in addition to artificial intelligence research, we are at the beginning of a wave of deep learning and application waves in the real world, involving speech recognition, image classification, object recognition and language in various industries. If the infrastructure and analysis parts of the ecosystem have evolved to the latter part, we are still very early pioneers for enterprise and vertical artificial intelligence applications.

Although the artificial intelligence start-up market can be said to have shown signs of eventual cooling, the start-up of deep learning-based start-ups continued one or two years ago. The overall size and valuation expectations are still high, but we are certainly going through a phase where large Internet companies will buy early artificial intelligence startups for talent. Compared with other companies that use this kind of hype, there are also some “real” artificial intelligence start-ups in the market. Some artificial intelligence start-ups that were established between 2014 and 2016 are beginning to take shape, with many companies offering increasingly interesting products across industries and verticals such as healthcare, finance, Industry 4.0 and back office automation. In the next few years, deep learning will continue to bring tremendous value to real-world applications, and artificial intelligence start-ups that focus on verticals will face many huge opportunities.

This continued explosion is largely a global phenomenon, with Canada, France, Germany, the United Kingdom and Israel being particularly active. However, China seems to be at a completely different level in terms of artificial intelligence. It is reported that the scale of government-led data collection is incredible (crossing Internet companies and municipalities), and the rapid identification of areas such as facial recognition and artificial intelligence chips. Development, and providing several rounds of huge financing for start-ups: According to CB Insights, China accounts for only 9% of global artificial intelligence transactions, but in 2017 global artificial intelligence funds accounted for nearly 48%, higher than 2016's 11 % (see some examples below).

Similarly, data privacy (and ownership and security) issues are becoming a major concern worldwide. In the early days of the Internet, data privacy was meant to protect what we did online, a relatively small part of our activities. Accordingly, only a small percentage of people really care about data privacy issues. As all aspects of our personal and professional life are connected to the Internet through more and more connected devices, the stakes are changing. Artificial intelligence can detect anomalies, predictive outcomes, and recognize faces in large data sets, making data privacy issues more complex.

Another independent but related issue is that many of these data are owned by the Large Internet Enterprise (GAFA). Some companies, such as Facebook, have proven to be not perfect managers. Nonetheless, these data provide an unfair advantage in their competition to produce more powerful artificial intelligence.

An emerging theme for these issues is to think of blockchain as a possible way to counter artificial intelligence risks, and as an alternative to producing better artificial intelligence in companies outside of GAFA. The crypto economy is seen as a way to motivate individuals to provide personal data and as a way for machine learning engineers to model these data by anonymizing it. All this is still in the experimental stage, but some early markets and networks are emerging.