Written by:Kang Xiang
Editor: Ayou
Design: Zicai
It is no exaggeration to say that among the many cloud services released by AWS, Amazon SageMaker absolutely belongs to the level of Star of the Stars. As a fully managed service, SageMaker helps developers and data scientists quickly build, train, and deploy machine learning (ML) models. How can one not love it?
Unfortunately, since its release in December 2017, Chinese users have been unable to access it—unless using an international AWS account—until May 12, 2020, when AWS officially announced that Amazon SageMaker was officially launched in the AWS China (Ningxia) region operated by West Cloud Data and the AWS China (Beijing) region operated by ChinaNet Center.
This is certainly great news! After all, implementing machine learning is an extremely complex task that requires not only professional skills but also involves a lot of trial and error. Whether it is ‘professional’ or ‘trial and error’, the costs behind them are high. However, how many companies can afford this? Many people may have a big question mark in their hearts.
Amazon SageMaker can eliminate the heavy work involved in various steps of the machine learning process. With pre-built common algorithms and automatic model tuning, Amazon SageMaker greatly reduces the difficulty of building and training models, allowing more enterprises to reap the benefits of machine learning.
Importantly, Amazon SageMaker significantly simplifies and accelerates the model training process, making machine learning more efficient. At the same time, with a recent series of important features and advanced capabilities, Amazon SageMaker enables customers to more easily build, train, tune, and deploy machine learning models.
At a recent media communication meeting, Dr. Zhang Xia, Chief Cloud Computing Strategy Consultant of AWS proudly stated that currently, tens of thousands of enterprises worldwide have chosen AWS to run machine learning workloads, at least twice as many as any other operator, including both traditional enterprises and a large number of innovative companies.
The ability to excel in machine learning is partly due to Amazon’s origins in e-commerce. From the beginning, Amazon has paid great attention to machine learning, applying machine learning technology extensively in warehousing, distribution, and many other work and business processes. After a long period of iteration, these technologies have naturally become more suited to application needs rather than just theoretical discussions.
Zhang emphasized that in the field of machine learning, Amazon does not work in isolation but fully supports various open-source frameworks. Different frameworks mean different characteristics and different usage scenarios, demonstrating that Amazon SageMaker adapts to the broadest range of production and application scenarios.
At the infrastructure level, Amazon’s virtual machines provide a variety of instances and ready-to-use Amazon Machine Images (AMIs). This means that customers can immediately start deploying machine learning by renting Amazon’s virtual machines, using various methods including containers.
The middle layer is the core service layer for machine learning, which includes the main products released by AWS this time, including Amazon SageMaker and Studio.
The top layer is the artificial intelligence service layer, which includes some dedicated services developed through deep learning, such as the speech-to-text service Amazon Polly and Amazon Transcribe, the cloud customer service solution Contact Lens, the recommendation engine Personalize, and the forecasting service Amazon Forecast, among many others.
Zhang stated that among the three layers mentioned above, the lower you go, the more foundational it is, requiring higher qualifications from users, but its functions are more powerful, enabling customers to do more things. The higher you go, the more specialized and specialized it becomes, so AWS also strives to productize and service it as much as possible to help more customers quickly use it.
According to Su Yingbin, Technical Director of Dayu Unlimited Machine Learning, the company is a user of AWS machine learning, primarily providing mobile short video services for emerging markets in Latin America, the Middle East, and Southeast Asia. The monthly active users of their main product Snaptube have exceeded 100 million.
He stated that the emergence of SageMaker helped Dayu Unlimited achieve a breakthrough from 0 to 1, greatly simplifying the entire process of building, training, and deploying machine learning. Many times, Dayu Unlimited is ready for training instantaneously, directly calling interfaces, setting parameters, and basically deploying online with just a few commands.
As a loyal user of AWS machine learning, Chen Changyou, Product Manager of Yikeluode also shared his experience.
According to him, Yikeluode is a cloud consulting service company, a core-level partner of AWS China, focusing 100% on AWS cloud computing, providing customers with cloud consulting services, migration, cloud hosting, etc., and will also assist them in optimizing architecture and cost structures.
Yikeluode can use Amazon SageMaker as a basis to build corresponding solutions for customers. For example, a marketing technology company can deeply analyze consumer data in the market through the solution provided by Yikeluode on AWS, providing insights and analysis to assist its clients in formulating marketing strategies.
Chen Changyou stated that similar work was not impossible in the past, but the workload was indeed too large. In addition to the existing hundreds of billions of transactions, new data continues to accumulate at a very fast pace, being diverse and inconsistent, making processing very difficult and the costs unbearable. However, through the AI model provided by Yikeluode, the system automatically analyzes and interprets a large amount of data, fundamentally solving the customer’s pain points.
Finally, Zhang stated that China is a country that quickly embraces innovation, and the popularization of various emerging applications in China is far faster than in developed countries like Europe and the United States. He believes that the landing of Amazon SageMaker will give more enterprises wings, making machine learning ubiquitous and ushering in a new era of innovation.
For more articles
Please long press the image below to scan the code to follow
Fat Head Tuo
