Kanta's Blog🇩🇪🇯🇵
Kanta Yamaoka
Personal Update: Recently joined @Google Japan

"Getting out of my comfort zone"; 1.5 years of working experience in an ML project with European colleagues @Rokken shaped my unique technical background. In my final undergraduate year, I stayed at German Research Center for Artificial Intelligence, @DFKI, one of the biggest world's non-profit research AI institutions from 2022 March to 2023 March.

Today's technology will be outdated tomorrow, but my learning ability keeps me adaptive to the rapidly changing world, as my professional working multilingualism, international experiences, and versatile computer science skills indicate. Hard work does not pay off from my bitter personal failure, so I always try to acquire skills in a deliberate manner: planning and practice.

I'm open to opportunities around the globe. Reach out via my LinkedIn . Let's have some coffee together if you're in Japan! (Virtual coffee is also great.)

Who am I?–From Japan to Germany.

I was born and raised in Japan. From March 2022 to March 2023, I lived in Germany, thanks to the following financial support: the JASSO study-abroad scholarship and the Research Assistant position at German Research Centre for Artificial Intelligence (DFKI).

Throughout my stay, I've worked on my bachelor's thesis at Osaka Prefecture University at DFKI/ TU Kaiserslautern in Germany, under the joint supervision of Prof. Dr. Prof. h.c. Andreas Dengel (Executive Director of DFKI), Jun.-Prof. Dr. Shoya Ishimaru, and Prof. Koichi Kise (Director of DFKI Lab Japan).

Publications—Int'l Conferences
[1] Kanta Yamaoka, Ko Watanabe, Koichi Kise, Andreas Dengel, and Shoya Ishimaru. 2023. Experience is the Best Teacher: Personalized Vocabulary Building Within the Context of Instagram Posts and Sentences from GPT-3.
In Proceedings of the 2022 ACM International Joint Conference on Pervasive and Ubiquitous Computing (UbiComp'22 Adjunct). Association for Computing Machinery, New York, NY, USA, 313–316. doi.org/10.1145/3544793.3560382   ResearchGate
[2] Kanta Yamaoka, Andrew Vargo, Motoi Iwata, and Koichi Kise. 2023. Towards Guided-Breath as an Intervention Within Computer-Assisted Learning.
In Proceedings of the 2022 ACM International Joint Conference on Pervasive and Ubiquitous Computing (Ubicomp'22 Adjunct). Association for Computing Machinery, New York, NY, USA, 136–137. doi.org/10.1145/3544793.3560335   ResearchGate
Publications—Domestic Conferences
[3] 野々宮 悠太, 柳 健大, 栗生 薫, 内田 時央, 岩下 雄一郎, 藤原 史規, 山岡 幹太, 石丸 翔也, 仲子 聡一郎, 岡村 浩史, 岩村 雅一, 黄瀬 浩一, 新谷歩. 集中治療室における短期的な死亡を予測する再帰型ニューラルネットワークを用いた動的予測モデルの提案:MIMIC-IVデータベースを利用した後方視的コホート研究. 第43回医療情報学連合大会 (第24回日本医療情報学会学術大会), 2023, to appear. [En] Yuta Nonomiya, Kenta Yanagi, Kaori Kuriu, Tokio Uchida, Yuichiro Iwashita, Fuminori Fujiwara, Kanta Yamaoka, Shoya Ishimaru, Soichiro Nakako, Hiroshi Okamura, Masakazu Iwamura, Koichi Kise, Ayumi Shintani. Recurrent neural network-based dynamic prediction model for short-term mortality in intensive care units: retrospective cohort study using the MIMIC-IV database. In Proceedings of the 43rd Joint Conference on Medical Informatics (the 24th conference of the Japan Association for Medical Informatics), JAMI, 2023 to appear.
[4] ImageToVocab: Finding New Words to Learn From One’s Personal Pictures and Their Generated Sentences.
This thesis, reviewed by Professor Koichi Kise in March 2023, is an extension of [1] and a summary of my research project at DFKI/RPTU under the supervision of Junior Professor Shoya Ishimaru. Some of the research has not been published, as I have already graduated from my university, but I hope to publish or make it publically available it in the future.
For example, if I import an image that I took in Rotterdam, Netherlands, from my own Instagram, ImageToVocab generates a text and shows it alongside the image. Each word on the screen is clickable, and learners can click on unknown words to find their definitions. This way, learners can find vocabulary words that are strongly tied to their own life experiences.
The latest version of the ImageToVocab is hosted and runs in a multicloud environment, leading to easily manageable and scalable user studies. The application is built on top of modern cloud infrastructure: Google Cloud (Cloud Storage, Firestore, Auth, Cloud Run, Cloud Function), OpenAI completion API (GPT, text-davinci-003), Instagram Basic Display API, Google Cloud's AI/ML services (Cloud Vision API, Natural Language API, Cloud Translate API—Today's Vertex AI), Python/TypeScript/JS backend (hybrid of Cloud Functions tied to Firestore events, FastAPI Python backend inside a Docker container hosted on Cloud Run), NextJS hosted on Vercel, New Relic for logging, and so on. In the future, I'd like to share how we built the system, if I have some energy left, haha.
Coffee with me?
Reach out via LinkedIn. I'm open to opportunities around the globe. Did you notice the blog's domain?—earth!
Research Experience
  • 2019 April - 2023 March. Completed my BEng at Osaka Prefecture University. (today's Osaka Metropolitan University) Advisor: Prof. Koichi Kise.
    • Thesis: "ImageToVocab: Finding New Words to Learn From One’s Personal Pictures and Their Generated Sentences."
    • I engaged research activities Intelligent Media Processing Group, which opened up an opportunity to study abroad at DFKI/RPTU with JASSO study-abroad scholarship.
  • 2021 March-2023 March: Research Assistant @ DFKI, German Research Center for Artificial Intelligence, supervisors: Prof. Andreas Dengel, and Jun.-Prof. Dr. Shoya Ishimaru
  • 2021 March-2023 March: Exchange Student @ TU Kaiserslautern, PsyberLab
  • Media Coverage
    My report for the research exchange program (Japanese) at DFKI/RPTU is available here. You can also find me in an article about a project between my university and DFKI. It was a side project, but I helped some Japanese students (some of whom are medical students) stay at DFKI and have discussions with researchers.
    Personal Posts