Hiroaki Bekku (3rd-year, Faculty of Environment and Information Studies) and Haruka Ozawa (2nd-year master's, Graduate School of Media and Governance) from Professor Jin Nakazawa's Laboratory Receive "Excellent Paper Award" and Other Awards at the 69th IPSJ Symposium on Ubiquitous Computing Systems (UBI)

At the 69th Symposium on Ubiquitous Computing Systems (UBI) of the Information Processing Society of Japan (IPSJ), a paper by Hiroaki Bekku (at the time of the award: 3rd-year student, Faculty of Environment and Information Studies; top photo) and others from Professor Jin Nakazawa's Laboratory, Faculty of Environment and Information Studies, received the "Excellent Paper Award," and a paper by Haruka Ozawa (at the time of the award: 2nd-year master's student, Graduate School of Media and Governance; bottom photo) and others received the "Student Encouragement Award."

The "Excellent Paper Award" is the highest honor, awarded to no more than one paper at each symposium. It is selected from all presented papers, regardless of whether the authors are professionals or students, and is judged on the content of the paper, including the oral presentation. The "Student Encouragement Award" is intended to promote student research. For papers submitted in the student category, no more than two are selected at each symposium based on a comprehensive evaluation of the quality of both the presentation and the paper.

[Excellent Paper Award]

Hiroaki Bekku (3rd-year, Faculty of Environment and Information Studies) and Haruka Ozawa (2nd-year master's student, Graduate School of Media and Governance)

"A Proposal of an Optimal Compressed Model Search Method Using Bayesian Optimization and Distillation"  

[Student Encouragement Award]

Haruka Ozawa (2nd-year master's student, Graduate School of Media and Governance) "An Automated Neural Network Construction Method by Controlling Accuracy, Inference Speed, and Power Consumption"  

I am honored to have received the "Excellent Paper Award" at the 69th Symposium on Ubiquitous Computing Systems (UBI) of the Information Processing Society of Japan. With the goal of building neural network models that run efficiently on limited computational resources, we used a compression method called distillation. Through a unique evaluation function and Bayesian optimization, we developed a method to search for the optimal model for compression tailored to specific objectives, such as "prioritizing inference speed" or "prioritizing accuracy." I am extremely happy to have received such a prestigious award as the "Excellent Paper Award" while still a third-year undergraduate student. I believe this was possible thanks to the cooperation of the professors, my peers, and the senior students in the Nakazawa Laboratory. In particular, I am very grateful to my senior and co-author, Haruka Ozawa. I would like to express my deepest gratitude. I intend to further develop this research and focus on creating a platform system for model compression, which we have named the "United Neural Network Compression Optimizer System (UNCO)."

I am delighted to have received the "Student Encouragement Award" at the 69th Symposium on Ubiquitous Computing Systems (UBI) of the Information Processing Society of Japan. In this research, we proposed a new method for automatically constructing neural networks, a foundational technology of AI. We proposed a method for automated neural network construction that takes the desired accuracy as input, controls for that accuracy, and considers inference speed and power consumption according to the application. Compared to existing research, our method achieves automated construction of neural networks corresponding to the desired accuracy without depending on set parameters. I also worked on research as a co-author with Hiroaki Bekku. Despite being only a third-year undergraduate, he is exceptionally talented, and I was very inspired by him. I am very proud that he won the "Excellent Paper Award" for his first-ever conference presentation, and I believe he will grow into an information technologist and researcher of global caliber. The reason I was able to receive this award is because of the support from everyone in the Nakazawa Laboratory; I feel that we received it as a team, not as an individual. Finally, I am deeply grateful to everyone in the Nakazawa Laboratory for always putting students first and providing a comfortable research environment.

Posted by: General Affairs Section, Shonan Fujisawa Campus (SFC) Office