"ModelArts is aptly named because it allows artificial intelligence modeling to become an art rather than hard work." — comment from a developer that test drove the platform during HUAWEI CONNECT 2018 in October of last year.
Zheng Yelai, President of the HUAWEI CLOUD BU, when releasing ModelArts (the faster AI development platform) had the following to say: "ModelArts is a full-process platform that assists developers with learning, training, and deployment. It is part of our Inclusive AI strategy, and this platform welcomes developers to join our ecosystem by lowering the threshold."
On January 30, 2019, HUAWEI CLOUD ModelArts was put into commercial use. The AI industry is developing rapidly. All industries have deployed the new accelerator in some form or another and many are jockeying to lead the innovations and further develop the technologies industries need to become more effective. However, for common enterprises and users, AI still presented a high threshold that few could cross. This should not be the case. Enterprises and developers, and even the platform providers, need to work together to overcome the challenges and promote AI adoptions. We need to come at the problems in AI together to grasp the technical dividends after scaling the walls.
What Are the Difficulties in AI Application Development? What Is the Ideal Development Platform?
Artificial intelligence is hot. Both enterprises and individuals want to share in the dividends but participating was no easy task before because it carried high initial investment. Previously, productivity was hindered by the cumbersomeness of managing exponential growth of data, increasingly complex models, time-consuming computing processes, and the host of development tools. The unwieldiness often squashed enthusiasm, which in turn quenched promotion of AI.
Faced with so many difficulties, many AI developers had to abandon their aspirations due to lack algorithm development and tuning capabilities. AI engineers, developers, and even novices can now quickly develop and deploy AI models and complete service development and rollout. ModelArts is precisely what everyone was waiting for in an AI development platform.
HUAWEI CLOUD ModelArts is meeting the requirements and slingshotting AI into a new sphere. As a one-stop AI development platform, it is adaptable to just about any use case with the needed software/hardware collaboration and device-side cloud integration to make it all happen. ModelArts provides modular-based services, such as automatic learning, data management, development management, model management, and reasoning service management. This pick-and-match flexibility enables developers to develop their AI models quickly and independently.
The Moxing high-performance distributed framework for AI algorithm engineers accelerates training for up to thousands of GPU clusters, often shortening the training period from weeks to minutes. ModelArts provides full-process AI development services such as massive data processing, large-scale distributed training, on-demand deployment of device-edge-cloud models, and O&M management. It helps AI developers quickly build and deploy models and manage the lifecycle of AI workflows, meeting the requirements of different development levels, reducing the AI development and usage threshold, and ensuring smooth, stable, and reliable running of systems.
For novices, AI models can be automatically generated to meet service requirements without any AI technology or code development capability. In addition, the platform can be used to produce the needed customizations for the use case.
Drilled-down development into AI to suit the particulars of the application just got much easier!
At the China International Big Data Industry Fair 2018, Zheng Yelai, President of HUAWEI CLOUD BU, said: "HUAWEI CLOUD is committed to lowering the technical threshold for AI and making the expertise of AI experts highly adaptable to the use case. All the complexities at the underlying end can be abstracted then further adapted in the overlay, thereby allowing AI to be tailored to the need instead of making the need fit into a rigid set of limited utilities. HUAWEI CLOUD is opening up these capabilities to everyone to enhance promotion of AI into industry practice. This approach has proven to be the right choice because without the wide-open inclusion, adoption would remain flat." Even those with limited knowledge of the tech can use ModelArts to create their own AI model with everything all in one spot so they don't have to grab from this and borrow from that and try to bring it all into the mix just to encounter some compatibility issue. This ease-of-use embodies part of the spirit behind the Inclusive AI strategy from HUAWEI CLOUD. This strategy is re-writing the rules so everyone can benefit, allowing everyone to participate in ways never before imaged.
The wide-open door of AI capabilities tears down the previous barriers of bogged-down efficiency; time-consuming labeling, training, and deployments; and the high expensive in procuring the necessary compute power to run the programs. For AI to scale in the verticals and achieve wide adoption, the threshold had to be lowered and the complexity in use has to be reduced. Developers have a lot of pain points, including such challenges as how to store and analyze at-scale data. Matters were further complicated for the average developer because they could not handle the rapid growth spurts in data sprawl.
Deep learning is the skill that developers must learn. However, for ordinary developers, the time cost of learning the deep learning model is quite high. To solve this problem, practitioners can use the available utilities in ModelArts to automatically adjust the parameters and yield the needed capabilities for the particular application, dramatically simplifying the workload of algorithm engineers so the model can be quickly converged into a viable and operable edition. For novices who do not have the ability to develop their own algorithms, ModelArts provides automatic learning utilities to help guide each step. The development process is accelerated through automations in data labeling, model training, and deployment, leaving the complex coding for the platform make the adjustments. Image classification, object detection, structured data prediction and analysis, sound and voice classification are just some of the support utilities available in the impressive platforming. ModelArts simplifies the process from beginning to end, delivering optimal results at affordable cost. That is, the platform automatically generates a model that meets the requirements based on the reasoning speed and finalized deployment environment.
In the future, ModelArts automatic learning will be implemented in various industries, covering multiple scenarios. For example, in water management projects, the operators can upload data to detect floats and help them automatically generate models based on automatic learning capabilities to monitor pollution.
In the construction industry, workers previously used outdated methods to inventory steel bars one at a time. This was slow and prone to error. Using the ModelArts image detection capability, the steel bars can be automatically identified, and the number of steel bars can be quickly calculated, yielding higher levels of accuracy with a shorter turnaround. Advanced applications, such as facial recognition, enhanced learning policies, enhanced learning controls, and similar are also available.
Beefed-up Performance for a Better UX
Users of all levels of exertise can use ModelArts to quickly develop their own AI application platforming. On March 20, 2019, Stanford University released the latest DAWNBench ranking. ModelArts was ranked first in image recognition training and inference (ResNet50-on-ImageNet; above 93% accuracy). It greatly reduces the model training time and achieves ultimate inference performance.
The launch of the ModelArts platform is an important element in building up the AI capabilities in the HUAWEI CLOUD ecosystem. In March 2019, HUAWEI CLOUD launched an AI marketplace for developers. The marketplace includes the first platform in China that allows developers to release and subscribe to AI model services. It aims to provide a secure, open sharing and transaction environment for scientific research institutes, AI application developers, solution integrators, enterprises, and privateers. The platform effectively connects participants in the development ecosystem to accelerate rollout of AI-related products.
Huawei is not taking its foot of the pedal either. The platform is benefiting from the homegrown Ascend 310 and 910 chips, giving developers the improved performance training and reasoning capabilities they need at a price they can afford. Huawei is continuing to integrate more data labeling tools to benefit more use case, such as automated driving, smart manufacturing, and smart city. ModelArts is the ultimate AI development platform for developers that is opening up the tangibiles for AI for Industry.
Those taking part in the development of the ModelArts platform at Huawei are rather proud of what it does and wanted the name to reflect the ease of using the model to create the masterpieces in AI adoption while giving developers the tools and adaptability they need to have mastery over their own model.
October 2018
Zheng Yelai, Vice President of Huawei and President of Cloud BU, unveiling the faster, full-process development tooling platform ModelArts at HUAWEI CONNECT 2018. Zheng was quoted as saying: ModelArts is a full-process platform that assists developers with learning, training, and deployment. The new platform promotes inclusion, extending an open invitation to developers by lowering the threshold in many substantial ways. It is a manifestation of our Inclusive AI strategy.
November 2018
ModelArts open beta test
December 2018
At the AICon 2018 Global Artificial Intelligence and Machine Learning Technology Conference, HUAWEI CLOUD organized a hands-on ModelArts workshop that attracted more than 100 developers to experience for themselves the amazing features of the one-stop AI development platform powered on the Enterprise Intelligence suite in HUAWEI CLOUD. Attendees also had the chance to rub shoulders with experts and put their skills up against other enthusiasts during the on-site competitions.
January 30, 2019
ModelArts enters full commercial use
March 2019
Stanford University rankings show ModelArts performed best in image recognition training and inference.
March 2019
The Hong Kong HUAWEI CLOUD Artificial Intelligence (AI) Developer Challenge was held. The event was jointly organized by HUAWEI CLOUD, Hong Kong Artificial Intelligence and Robot Society (HKSAIR), and Hong Kong Science and Technology Park Company (HKSTP). Participants wrote the coding and developed AI applications right on the HUAWEI CLOUD ModelArts platform then offered their oral defenses for their innovations.
March 2019
HUAWEI CLOUD officially released the first AI model marketplace in China. As a developer ecosystem built on the one-stop AI development platform, the AI marketplace provides the modeling, API transaction execution and data collection, and references so research institutes, AI application developers, solution integrators, enterprises, and privateers can share their knowledge and cases in a secure and open environment to accelerate development and implementation of AI-related products.
April 2019
HUAWEI CLOUD Helps RFCx develop an innovative platform that integrates information collection devices, storage services, and intelligent analysis systems. An intelligent algorithm model is developed to accurately identify and differentiate the noises emitted by electric saws and trucks with the capabilities of the artificial intelligence services and the ModelArt tooling on HUAWEI CLOUD.
May 2019
On May 7, the 2019 Digital China Innovation Contest was launched in Fuzhou. The event carried the theme "Software Enablement Benefiting Innovations in the Economy to Drive the Digital China Agenda". As one of the sponsors, HUAWEI CLOUD presented the topic on "Cultural Heritage - Identifying the Intricacies in Chinese Calligraphy" that was able to use the tooling to pick up on characters that were seemingly impossible to distinguish. ModelsArts dazzled the crowd and many walked away amazed after experiencing it for themselves.
May 2019
On May 9, 2016, the results of the product awards selection for the 2019 AI Expo were announced. The HUAWEI CLOUD one-stop AI development platform ModelArts won the Gold Award.
