Ueda Research Laboratory
Ueda Research Laboratory
- Our paper, “Read the Silence: Well-timed Recommendation via Admixture Marked Point Process,” was accepted by AAAI-17 (AAAI Conference on Artificial Intelligence) in November 2016.
- Our paper, “A Semi-supervised AUC Optimization Method with Generative Models,” was accepted by ICDM2016 (IEEE International Conference on Data Mining) in September 2016.
- Two of our papers, “Higher-Order Factorization Machines” and “Multi-view anomaly detection via robust probabilistic latent variable models,” were accepted by NIPS2016 (Neural Information Processing Systems) in August 2016.
- The NTT fellow, Dr. Ueda, received an Achievement Reward from the Institute of Electronics, Information and Communication Engineers for "Pioneering Study on Statistical Machine Learning."
- Our paper, “Polynomial Networks and Factorization Machines: New Insights and Efficient Training Algorithms,” was published in ICML2016 (International Conference on Machine Learning) in June 2016.
- Ueda Research Laboratory was established in April 2016.
Spatio-temporal multidimensional collective data analysis
We aim to contribute to the realization of NTT's "ambient artificial intelligence" for real-time predicting "when" "where" "what" and "what will happen", and to develop a proactive control/navigation technology that gives feedback to social systems. We recently proposed new approaches to proactive optimal group navigation that involved observation, then prediction, and then simulation done repeatedly for proactive people flow navigation. Now, we plan to work on behavior principles such as those of people flow that are unknown using learning-type multi-agent simulation.
Sign discovery, causal relation inference
We aim to contribute to solving major problems related to NTT's businesses such as discovering signs of network failure by establishing technologies that detect signs of abnormal phenomena in real time from a variety of sensor data and log data. Obtaining one-to-one supervised learning data with stream data (such as sensor data) is difficult, so conventional supervised learning methods are hard to apply. We are planning to expand to applications such as predicting signs of abnormality and identifying behavior by establishing technologies for detecting object classes (events) instantly from stream data.
New developments in statistical machine learning
We aim to contribute to the development of natural sciences and social sciences with an aim on new developments in machine learning research in uncharted fields in big data analysis, such as big data from space (JST CREST) and smart cities (Ministry of Internal Affairs and Communications NICT fund). For example, for big data from space (JST CREST), we aim to be a pioneer in new areas of study (statistical computation for the study of physics), focusing on the world’s only analysis of space imaging data in cooperation with the University of Tokyo Institutes for Advanced Studies, Kavli Institute for the Physics and Mathematics of the Universe.
We are looking for a talented postdoctoral researcher in one of the following topics:
- machine learning
- mathematical optimization
- spatio-temporal analysis
- PhD in one of the above or related fields
- Proven track record of publications
- Good programming skills
- English fluency
Ability to write and speak Japanese is preferred but not required.
- Conduct innovative research in one of the above fields
- Present research findings both internally and at conferences or journals
- Engage in team collaborations to meet research goals
- One year with possibility to renew
- Competitive salary
Our group, located in Kyoto, researches the foundations of machine learning. We publish regularly in top-tier machine learning conferences including ICML, NIPS, AISTATS, KDD and others.
How to apply
Candidates should apply by sending an email including:
- a CV
- a research statement (one page)
- two publications
Start date should be spring 2017. Exact date is negotiable. Candidates will be reviewed on a regular basis until the position is filled.
Please send applications or inquiries to Naonori Ueda (firstname.lastname@example.org)