Amazon cloud services (AWS) expands the circle of machine learning to reach every AI worker
RecordTrend.com is a website that focuses on future technologies, markets and user trends. We are responsible for collecting the latest research data, authority data, industry research and analysis reports. We are committed to becoming a data and report sharing platform for professionals and decision makers. We look forward to working with you to record the development trends of today’s economy, technology, industrial chain and business model.Welcome to follow, comment and bookmark us, and hope to share the future with you, and look forward to your success with our help.
The following is the Amazon cloud services (AWS) expands the circle of machine learning to reach every AI worker recommended by recordtrend.com. And this article belongs to the classification: Industry information.
On December 9, Swami sivasubramanian, vice president, global machine learning, AWS (Swami for short) delivered a keynote speech on machine learning and artificial intelligence at Amazon re: invent conference, showing the latest panoramic blueprint of AWS on artificial intelligence and machine learning, and announced a series of new services and functions, making machine learning easier to use and expand to a wider range of users, application scenarios and industries. This is the first machine learning keynote speech at Amazon re: invent. “Machine learning is one of the most disruptive technologies that our generation can encounter,” Swami said in a keynote address. “At present, more than 100000 customers are using AWS’s machine learning service, and many customers have used machine learning in their core business.”
Gu fan, general manager of AWS cloud service product management in Greater China, said, “Amazon has been using machine learning technology for more than 20 years, which is the deep source of AWS machine learning service. AWS started to provide machine learning services on the cloud in 2016. Only three services were released in that year, and it began to accelerate in 2017. In the last three years, more than 200 new services and functions were added each year, enriching the tool set urgently needed by the global AI workers. “
According to the white paper on global AI Development released by Deloitte, the world AI market will exceed US $6 trillion by 2025.
In the face of the development opportunities of digital economy, many countries and regions have listed artificial intelligence as the national strategy of priority development.
On November 21, 2020, the National Industrial Information Security Development Research Center pointed out many difficulties in the integration of artificial intelligence and manufacturing industry in 2020 white paper, in which the problem of talent shortage is particularly serious. According to the report on the official website of the Ministry of human resources and social security, the gap of artificial intelligence talents in China is more than 5 million, and the domestic supply-demand ratio is 1:10, which is seriously unbalanced.
According to the white paper on the development of global artificial intelligence published by Deloitte, machine learning is one of the core research fields of artificial intelligence. 89% of AI patent applications and 40% of AI related patents are machine learning.
AWS is a leader in cloud computing and a leader in machine learning. In the face of machine learning, a promising enterprise, and the current situation of serious shortage of talents, AWS adopts a series of measures through various ways, focusing on the expansion of machine learning circle through technological innovation.
First out of the box machine learning solutions for Industry
One of the measures to expand the circle is to launch out of the box solutions. At the re: invent conference, AWS released five machine learning services for industry, namely Amazon monitron, Amazon lookout for equipment, AWS panorama all-in-one machine, AWS panorama SDK and Amazon lookout for vision. This is the first time that AWS has launched out of the box machine learning solutions for industrial fields.
Amazon monitron and Amazon lookout for equipment support predictive maintenance through machine learning. Amazon monitron provides an end-to-end machine monitoring system consisting of sensors, gateways and machine learning services for customers without sensor networks to detect anomalies and predict when industrial equipment needs to be maintained. Amazon lookout for equipment is aimed at customers who already have sensors but do not want to build their own machine learning models. AWS builds models for them and returns prediction results to detect abnormal device behaviors.
AWS panorama improves industrial operations and workplace safety through computer vision. AWS panorama all-in-one machine is a hardware device, which can automatically identify the camera data stream and interact with the industrial camera by connecting it to the network of the industrial site. The AWS panorama software development kit (SDK) makes it easy for industrial camera manufacturers to embed computer vision into new cameras.
Amazon lookout for vision provides industrial customers with high-precision and low-cost solutions for product quality anomaly detection. Through machine learning technology, it can process thousands of images per hour, and find product defects and anomalies. Customers can send camera images to Amazon lookout for vision in batches or in real time to find out anomalies, such as cracks in machine parts, dents on panels, irregular shapes or color errors on products.
At present, customers and partners who have already used machine learning services in the industrial field of AWS include axis, Linghua technology, BP, Deloitte, fender fanda, GE Medical and Siemens transportation, etc.
Build an all inclusive toolbox and empower every AI worker
The second measure to expand the circle is to create a comprehensive and rich tool set, in Gu fan’s words, right tool for the right job. The machine learning tool set provided by AWS includes three levels.
At the bottom of the toolset, for those customers with strong technical ability, they hope to take artificial intelligence and machine learning as their core competitiveness. AWS provides them with powerful computing power, comprehensive selection of computing power and rich machine learning framework. AWS supports mainstream machine learning frameworks, and customers can also bring their own machine learning frameworks through container deployment; AWS can provide powerful computing power based on the latest processors of NVIDIA, Intel, AMD, and Xilinx, and greatly reduce the computing power cost of machine learning through self-designed processors.
The middle layer of the toolset is for customers with strong technical ability. They have a large amount of data for machine learning model training. They have certain algorithm talents. They don’t need to spend energy on infrastructure management and focus on their own application and business innovation. Amazon of AWS Sagemaker provides them with the first fully managed machine learning integrated development environment, and continuously adds new functions to the development environment, from data preparation, to model training, parameter tuning and model iteration, to model deployment and model quality monitoring, in the whole process, it maximizes the efficiency of machine learning and reduces the threshold for them to carry out machine learning.
The top level of the toolset is for customers with relatively weak technical ability. They have certain data but no algorithm talents. They hope to introduce artificial intelligence directly into business scenarios. AWS provides them with out of the box AI services, including machine vision, voice and text conversion, machine dialogue, text processing, e-commerce business, customer service, enterprise information search, development and operation and maintenance, industrial AI, etc.
Through such a comprehensive tool set, AWS can cover and empower all AI workers.
Expand to database developers and data analysts
The third is to extend machine learning to data developers and data analysts. The number of database developers and data analysts is much larger than that of machine learning developers. They do not have the knowledge and skills of machine learning, but they do not lack the idea of machine learning. Therefore, AWS grafted the machine learning ability with the database, allowing database developers and data analysts to follow the way of database query, so that their machine learning ideas can be applied to business applications. Amazon aurora is a famous relational database service of AWS. AWS has launched a new function Amazon Aurora ml for Aurora. When a database developer initiates a database query (SQL), as long as a machine learning model is selected, the machine learning service will be awakened. Aurora ML will automatically give the query results to the machine learning model for reasoning and return the results. For example, to query whether a customer’s evaluation is positive or negative, the database developer only needs to query the database and select the model, and the returned query results will automatically add positive or negative judgments. Similarly, the offshore e-commerce company wants to change the commodity information in the database into multilingual. The database developer only needs to query the commodity information and select multilingual translation, and the returned results will automatically include the multilingual translation of the commodity information.
Amazon Athena is a service often used by data analysts. Through this service, you can use SQL statements to query data directly from the object files on Amazon S3 (SQL is a structured query language, originally used for relational data query, but the object file of S3 is not relational data). AWS also launched a new function Amazon Athena ml, query results can be automatically attached to the results of machine learning reasoning.
Amazon redshift is a cloud native data warehouse. Amazon redshift ml, a new function launched by AWS, even saves the step of selecting model. For example, in the field of e-commerce, which customers are likely to lose? At this time, you may not have a model to determine what characteristics of customers may be lost. Through redshift ml, data analysts only focus on SQL queries. Redshift ml can import data into S3, and then combine the autopilot function of sagemaker. Autopilot is an automatic modeling function. Such redshift ml can automatically carry out data cleaning, model training, and select the best model for prediction.
Amazon Neptune is a graph database of AWS, which is mainly used for knowledge mapping, identity mapping, fraud detection, recommendation engine, social relations, life science and other scenarios. It uses graph to represent the relationship between various data entities, such as friend relationship graph. For graph database, it is obviously not enough to just show the correlation of data. What users need more is machine learning and reasoning according to these correlations. The new function Neptune ml is to connect graph database with machine learning, visit graph database through machine learning model, and make more accurate prediction.
Amazon quicksight is a business intelligence (BI) service of AWS, which can easily call various data for analysis and presentation. AWS launched quicksight ml in May 2020, which is also combined with the autopilot function of sagemaker. Data analysts can use it to carry out fraud detection and sales forecasting.
At this year’s Re: invent conference, AWS launched a cooler new machine learning feature quicksight Q. Through it, we can use natural language to ask questions about data and get the data insight we want. For example, enter “what is our year-on-year growth rate?” directly into the query box You can get a highly accurate answer in seconds. If the growth rate needs to be pre-defined in the model, update the model, and process the data, it may take days or even weeks.
AWS also launched Amazon lookout for metrics, which uses machine learning technology to detect data anomalies by comparing multiple enterprise data. Gu fan, for example, said that the selling price of a commodity is 200 yuan, which becomes 20 yuan in a certain data source. It is of great significance to find such abnormal data through Amazon lookout for metrics. If such a price error occurs in online sales, it may bring huge losses to the enterprise.
In addition, AWS also released Amazon Devops guru, an operation and maintenance service using machine learning, which can help application developers automatically detect operation and maintenance problems, give remedial measures and improve application availability. Previously, AWS has launched Amazon CodeGuru, which allows developers to use machine learning to automate code auditing and provide guidance and advice.
Amazon sagemaker adds nine new features to make it faster, simpler and simpler
Fourth, we should vigorously develop the intermediate force of machine learning. As mentioned earlier, Amazon sagemaker is an integrated development environment for machine learning developers and a fully hosted service. It eliminates the challenges in each stage of the machine learning process, simplifies the complexity, and enables developers and data scientists to build, train and deploy machine learning models more easily and quickly. Amazon sagemaker’s capabilities are also in rapid iteration, delivering more than 50 new features in the past year. At this year’s Re: invent conference, AWS released nine new features again.
(1) Data wranger, data feature extractor. Amazon sagemaker data Wrangler simplifies data preparation for machine learning. One of the most important tasks in machine learning training, called feature engineering, is to extract data from data of different sources and formats to form standardized data fields (also known as features) as the input of machine learning model. This work is very time-consuming. Through data Wrangler, customers can import data from various data stores with one click. With more than 300 built-in data converters, data Wrangler allows customers to standardize, transform and combine features used by machine learning without writing any code. Customers can quickly preview and check whether they meet expectations by viewing these transformations in sagemaker studio, the first end-to-end integrated development environment for machine learning.
(2) Feature store, data feature repository. Given the large number of features to manage, AWS has introduced a new feature for Amazon sagemaker called feature store. It is a special library for updating, retrieving and sharing machine learning features. After the features are designed through data Wrangler, they can be saved in the feature store for reuse. A set of features will be used in different models and used by multiple developers and data scientists. It is necessary to track and manage these features effectively, update them in time, and maintain consistency. Model training and reasoning by using model (that is to say, using model in practice) have different usage scenarios for features. In the process of training, the model can access features offline and in batches, and it takes a long time to use. For reasoning, only a part of the feature base is usually used, but real-time access is needed, and the prediction results are returned within a few milliseconds. Therefore, how to manage the feature library is a complex matter, and feature store is used to solve these problems.
(3) Pipelines, automated workflow. Like traditional programming, choreography and automation can improve the efficiency of machine learning. Amazon sagemaker pipelines is the first user-friendly CI / CD (continuous integration and continuous delivery) service built for machine learning.
(4) Clarify, model deviation detection. Through Amazon sagemaker clarify, developers can easily detect the statistical deviation in the whole machine learning workflow, explain the prediction made by the machine learning model, identify the deviation, clearly describe the possible sources and severity of the deviation, and guide the developers to take measures to reduce the deviation.
(5) Deep profiling for Amazon sagemaker debugger. Through deep profiling, it can automatically monitor the system resource utilization, such as GPU, CPU, network throughput and memory I / O, and alarm the resource bottleneck in the training process, so that developers can schedule resources in time and train the model faster.
(6-7) distributed training, distributed training of large complex deep learning model. AWS provides two methods: model training is split into hundreds or thousands of CPUs. One is the data parallel engine, which splits the data set. One is the model parallel engine, which is the best way to automatically analyze and identify the segmentation model, and efficiently segment large complex models with several billion parameters on multiple GPUs. By splitting the training, Amazon sagemaker can train large and complex deep learning models twice as fast as the current method.
(8) Edge manager, edge model quality monitoring and management. Amazon sagemaker edge manager can help developers optimize, protect, monitor and maintain machine learning models deployed on edge device clusters. After models are deployed to edge devices, they still need to be managed and monitored to ensure that they continue to run with high precision. When the accuracy of the model decreases over time, developers can retrain the model and continuously improve the quality of the model.
(9) Jumpstart, a fast start tool. With Amazon sagemaker jumpstart, customers can quickly find information about machine learning scenarios similar to their own. Novice developers can choose from a number of complete solutions, such as fraud detection, customer churn prediction or time series prediction, and deploy them directly into their Amazon sagemaker studio environment. Some experienced users can choose from more than 100 machine learning models to quickly start model construction and training.
Amazon sagemaker is popular with customers because of its rich new features. In just three years, it has been used by tens of thousands of customers, including 3M, ADP, AstraZeneca, Avis, Bayer, Bundesliga and capital One, Cerner, Chick-fil-A, harmony, damelor Pisa, Fidelity Investment, GE Healthcare, Georgia Pacific, hurst, iFood, iheartmedia, JP Morgan, intuit, Lenovo, LYFT, NFL, nerdwallet, T-Mobile, Thomson Reuters, vanguard, etc.
Behind the series of moves to expand the circle of AWS is the ambition of AWS for machine learning. Swami said he graduated as a graduate student 15 years ago and was fortunate enough to enter AWS to start cloud computing. Today, it is no exaggeration to say that cloud computing has released great power and helped various start-ups and mature enterprises to achieve great success. Machine learning is now in that early stage. As we can read between the lines in swami, machine learning is the next gold mine for AWS.
About Amazon re: invent
Since 2012, Amazon re: invent is the annual event held by Amazon cloud services (AWS), the global leader in cloud computing, as well as a comprehensive and grand industry summit in the field of global cloud computing. Every year, Amazon re: invent publishes a series of innovative technologies and services that will lead the future, inviting customers and AWS partners from all walks of life in the world to share the latest business innovation practices. As a result, it has become a weathervane of the cloud computing industry, attracting wide attention and participation of developers and users all over the world.
“Top technology, reshaping the future.” Amazon re: invent 2020 is ready to go. The 3-week online Summit (December 1-december 18, 2020) is open to the public free of charge for the first time. Welcome to: https://reinvent.awsevents.cn/ Watch the agenda.
Read more: Amazon cloud services (AWS) speeds up the landing of cloud products and services in China from the professional field to the mass sector. AWS joins hands with China’s local travel giant Shouqi to launch five machine learning services for industrial use AWS releases Amazon Devops guru machine learning driven new operation service AWS Zhang Xia: tensorflow 85% of the global load is on the AWS platform Amazon sagemaker Amazon cloud service (AWS) launched in Ningxia and Beijing region of China Amazon sagemaker promotes innovative application of machine learning It is expected that more than 50 technical forums and more than 6000 professionals will attend the meeting, which will be the highest in China in the AWS technology summit. AWS will release financial management services in China to make customers more economical and efficient use of cloud services. AWS will release a new generation of Amazon Aurora serverless, which is more convenient for SQL server migration Sagemaker Amazon cloud service (AWS) helps Toyota Internet China’s Internet of vehicles service to be fully implemented. BMW Group and AWS work together to drive automotive industry innovation with data. Tom cat game family of 40 million daily lives creates “insensible” user experience with AWS cloud service
If you want to get the full report, you can contact us by leaving us the comment. If you think the information here might be helpful to others, please actively share it. If you want others to see your attitude towards this report, please actively comment and discuss it. Please stay tuned to us, we will keep updating as much as possible to record future development trends.
