Growing Together with MindSpore: Zhenbang's AI Exploration Journey

Agriculture, as the cornerstone of humanity's survival and development, is still facing difficulties such as water scarcity, irrational land use, and low productivity. With the surge of science and technology, AI technologies are developing rapidly and have been widely used in various fields, providing new ideas for solving these agricultural challenges. How to use AI technologies to improve agricultural productivity, utilize agricultural resources rationally, and improve the living quality of farmers has become a hot topic.

Wang Zhenbang, a student at Wuhan University of Technology, has been dedicated to agricultural issues for a long time. In the 8th "Internet+" Competition, he used MindSpore AI technologies to facilitate plant breeding and promote the transformation and upgrade of agriculture. With the power of MindSpore in hand, Zhenbang has overcome various challenges. Now, let's follow his steps and see how he advanced on his AI development path.

01 A Millennial Developer Committed to Self-Developed Solutions

In this era full of opportunities and challenges, young developers show unlimited creativity and passion and inject new vitality into all walks of life. Supported by the guidance of his mentor, Zhenbang joined the MindSpore community and started his AI exploration. From simple operator development to model migration, he started from scratch and learned while doing during the development of each project.

After accumulating a large amount of model experience in various fields, he started to summarize and integrate models in each field to form a suite for the current field, discovering errors during forward advancement and back-propagating to optimize the previously developed operators and models. Over time, Zhenbang's optimization promotes the development of the community, helping MindSpore grow stronger.

During development and training, Zhenbang set his goal as "to develop an easy-to-use AI framework that belongs to our community."

02 Growing Together with MindSpore

"It's like the first flap of the butterfly's wings. It's a strong recognition that gives me great courage to advance in the journey of MindSpore." In a MindSpore community program, community experts invited Zhenbang to share his experience among university developer representatives. This was not only a simple sharing, but also an opportunity and an important node that gave him courage and strength to try and participate in a variety of projects in the community.

There are many opportunities for new developers in the MindSpore community. The MindSpore community has built an end-to-end talent development system that covers learning, training, practice, and competition. It provides developers with practice courses from entry-level to professional, initiates multiple project competitions and open source internships, and carries out the OpenMind project with universities, providing comprehensive project practice opportunities. Excellent developers are invited to share their experience in the MindSpore Study Group (MSG), MindSpore SIG, and Ascend AI Developer Summit. Whether you want to study algorithm applications, optimize underlying code for high performance, or share project experience, you can always find appropriate projects in the community for in-depth research.

Cultivated by Professor Xiong Shengwu, Dean of the School of Computer and Artificial Intelligence, WHUT, Zhenbang grew from a team member and leader of the CPU operator development team to the CPU operator development owner and then to a MindSpore evangelist. He participated in nine MindSpore projects, contributed 15,774 lines of code, and developed 48 GPU operators. The OpenMind project enables him to grow rapidly.

Now, Zhenbang, a MindSpore evangelist, acts as a developer, user, and promoter of MindSpore. He submits pull requests to fix or optimize the MindSpore framework and shares the technical highlights with others. He is proud to participate in such a large open source project and help the community solve problems in the framework.

We are glad to see that the young man with a sense of mission has slowly grown up, and he will go higher and further with bravery and enthusiasm.

03 Facilitating Agricultural Upgrade with AI

After accumulating some development experience, Zhenbang turned his focus to the development of the physical industry. Under the guidance of postdoctoral researcher Rong Yi from the Yazhou Bay Seed Laboratory, Zhenbang began the research on AI-assisted plant breeding. He found that China's agricultural field mainly relies on artificial breeding, while the international trend has developed towards the era of breeding 4.0, that is, biotechnology + AI + IT. Knowing his technical knowledge could be applied to industrial development, Zhenbang found his goal. In the "Internet+" competition, he chose to use AI to develop breeding technologies. He said, "Seeds are chips in agriculture. My goal is to cultivate our own seeds."

During development, Zhenbang's team mainly uses the model repository of MindSpore to quickly build models and uses few-shot learning and Transformer technologies to process time sequence information and capture important and representative loci in the gene sequence to complete the construction of the entire model. The finished model of his team delivers 16% higher prediction accuracy than traditional breeding and gene prediction technologies, as well as significantly higher inference speed and data throughput due to model deployment on MindSpore. Zhenbang's team is undergoing long-term cooperation with BGI Group to continue polishing the project into a feasible breeding system.

04 Conclusion

As a young, excellent MindSpore developer, Zhenbang has developed basic operators and models, and then used AI technologies to provide a powerful driving force in the agricultural field. From his growth journey, we can see the enthusiasm of next-generation developers. MindSpore has been and will be accompanying the young developers to improve the community mechanism and build a MindSpore AI learning and practice platform. MindSpore will grow with the developers.