Google is collaborating with a student team at Purdue University to create open-source TensorFlow models. TensorFlow is a deep learning framework created by Google. The student team will implement recently-published algorithms and methods in machine learning. The implementations will be made available through a public repository called “Model Garden” maintained by researchers and engineers from Google.

Jaeyoun Kim, a Technical Program Manager at Google, said, “Machine learning is a fast-growing field, and there are often different machine learning model implementations available for the same algorithm. As a result, it is not easy to distinguish which one is high-quality and reproducible. The concept of a Model Garden is to ensure that good models flourish without weeds.”

This Google-Purdue collaboration will pioneer a new model of collaboration. A Vertically Integrated Projects (VIP) team will train Purdue students to write TensorFlow model implementations. The students can continue taking courses while gaining valuable experience interacting with Google experts in machine learning. Upon completing a project successfully, Purdue students will have gained valuable skills that make them more competitive for jobs at companies using machine learning, such as Google. 

VIP is a project-based course; each team is advised by a faculty member (or co-advised by several faculty members). Vertical integration means students of all years, from first-year to graduate students, can participate. “Such integration is similar to what people encounter in the real workplace. VIP experience helps students understand how to collaborate with people of different backgrounds, knowledge, experience, and skills.”, said Dr. Carla Zoltowski, the director of VIP.

Many VIP teams have sponsors. However, this team adopts a new approach: a process of two stages for training and hiring. Students interested in participating in this project can visit the VIP’s website (https://engineering.purdue.edu/VIP/teams) for more details.

This team will be advised by Dr. Yung-Hsiang Lu, a professor of Electrical and Computer Engineering and a member of Purdue’s Center for Programming Principles and Software Systems. “This team will offer a unique learning experience because in most courses students’ work is thrown away after grading. In contrast, students in this team will help create a good foundation for researchers worldwide.”, said Dr. Lu.