March 28, Tencent announced the launch of DI-X depth learning platform, and further open Tencent in the field of leading-edge capabilities in the AI, help enterprises into the artificial intelligence fast lane.
With the release of DI-X platform, Tencent cloud in the field of AI's full range of layout also surfaced: from IaaS infrastructure, AI platform services, AI basic services, AI application services, to the vertical industry solutions, have perfect products And service coverage. Ma has said: "artificial intelligence, things, and even the future of unmanned, robots, etc., its core of the background must have a brain in the clouds." "DI-X platform for the introduction of enterprises to build" cloud on the brain "to provide the most powerful help, this also announced Tencent cloud in the AI layout of the overall speed.
DI-X depth learning platform on the line
DI-X platform to help enterprises into the artificial intelligence fast lane
Based on Tencent Cloud's powerful data storage and computing power, the DI-X platform is a one-stop, deep learning platform for development, training, forecasting and deployment. It is suitable for image recognition, speech recognition, natural language processing, machine vision And other popular areas. The opening of the DI-X platform allows algorithm engineers and data scientists not to focus on machine learning, especially in the depth of learning the underlying engineering cumbersome details and resource constraints, while focusing on model and algorithm tuning.
With Tencent cloud many advantages of products and capabilities, DI-X platform has experienced from the inside out of the output process. With the development of machine learning and artificial intelligence, Tencent needs a platform to support the urgent needs of internal algorithm engineers and data scientists, DI-X came into being, and become the main machine learning platform, every day tens of thousands of machine learning task flow, On top of them continue to run a variety of algorithms and train a variety of models. After more than a year of actual combat. With the system mature and external demand increases, Tencent cloud decided to open this ability.
Compared with other depth learning platforms in the industry, DI-X platform has many improvements and innovations, in the ease of use, flexibility, compatibility and other aspects of outstanding performance. Such as DI-X of the depth of learning components are the latest open source version of the industry, get through the interaction with the object storage COS, multi-card multi-card acceleration, with Angel will be able to provide more performance support. To design the example, DI-X in the model of the collection, use, forecast, deployment has more advantages, the depth of learning more friendly.
With the DI-X platform as a bridge, users can build a variety of data, including pictures, voice and video, previously stored in the COS, through the GPU cloud server, easy to learn the depth of the various algorithms to build their own data quickly , Into model productivity, access to the AI fast lane.
"Cloud on the brain & rdquo; Accelerate the application of artificial intelligence scene landing
Tencent cloud in the field of AI's full range of layout, including IaaS infrastructure, AI platform services, AI basic services, AI application services, and vertical industry solutions. In addition to comprehensive product and service layout, Tencent cloud AI's unique advantages include a complete LBS intelligent services, data with the ability to recommend intelligent services, support for FPGA cloud server, and these capabilities have been Tencent massive products and services to verify.
AI layout capability matrix
Taking DI-X platform as an example, we introduce the algorithms and models for social, advertising, game, video and resource platforms, such as precision recommendation, relationship chain, game behavior insight, personalized delivery, advertising user portraits, These capabilities have been widely available, mobile phone housekeeper, application treasure, every day Express and other products have been validated, business users can quickly access and use.
In the scenario of advertising click behavior prediction, with the DI-X platform, a BRNN Encoder model (bidirectional cyclic neural network encoder) can be easily dragged out to extract the basic characteristics from the behavior sequence data of the user's own and user circle friends, (Stacked Auto-Encoder) model, and make full use of the model characteristics of RNN, and obtain the more accurate behavior prediction effect than the conventional model.
Image recognition judgment scene, DI-X provides Faster R-CNN and other image recognition algorithms, and support the convenience of ultra-parameter tuning, training model can be directly deployed for the new picture to predict, judge The basic properties of the items in the image. Identification results can be widely used in the field of public security network, advertising recommended areas of new product identification, intelligent transportation field of vehicle identification and other scenes.
Not long ago, Tencent AI Lab independent research and development of artificial intelligence scene application "jiorka" Go robot won the UEC Cup World Computer Go Conference champion, on behalf of Tencent AI technology strength has entered the forefront of the world. Tencent in the field of deep accumulation of artificial intelligence will continue through Tencent cloud open. At the end of last year, Tencent cloud for the global business launched breath detection face, facial features, face matching and verification, face search, picture labels, ID card OCR identification, business card OCR recognition 7 AI cloud services, promote AI Technology from hyun technology to achieve landing applications. In January of this year, Tencent cloud pioneered the introduction of the domestic first high-performance heterogeneous computing infrastructure and cloud cloud server, cloud services to large companies to long-term use of the FPGA to promote the use of more enterprises. The first half of this year, Tencent cloud will also be based on NVIDIA's latest Pascal architecture 1 8-card GPU server, single-machine multi-card machine performance is higher, to meet the large amount of data large-scale machine learning algorithm needs to maximize the efficiency of computing The
With the boom of artificial intelligence, the major Internet giants are accelerating the layout, and AI has gradually become an important ability to accelerate the expansion of ecological areas. The opening of the DI-X platform will help enterprises to quickly get the strongest "cloud of the brain," and quickly cut into the AI's fast lane, and Tencent cloud comprehensive AI capacity layout will also enhance the enterprise in the artificial intelligence scene application of the ability to advance Wave of wave technology in the occupation of the initiative.