CyberAgent

Position Title/Short Description

Full Text of JOE Listing:

About CyberAgent:
CyberAgent has been expanding its business in Japan based on the rapidly growing internet industry since its founding with the vision of “To create the 21st century’s leading company". We operate “ABEMA”, a new future of television, as well as an Internet advertising business that boasts the top share in Japan. CyberAgent continues to expand its business while creating new businesses to keep up with changes in the Internet industry.

About AI tech studio and AI Lab:
AI tech studio is a division of CyberAgent that develops cutting-edge services in the field of digital marketing. Currently, we are applying the product development process know-how we have cultivated for a long time in digital advertising to areas such as retail, healthcare, and digital government.

AI Lab, a research and development organization of AI tech studio, focuses on the area related to digital marketing. AI Lab is committed to advancing fundamental research in AI as well as making a broad impact on society through our service. Since AI Lab is not only providing research solution to the existing services and also providing core technologies in making new services, there are many researchers from variety of fields, such as machine learning, econometrics, market/mechanism design, computer vision, natural language processing, and human-computer interaction.

In addition, we collaborate with academic institutions to solve the problems our services face everyday in an academically valuable way and also bring state-of-the-art research outcomes to the core business value in our service.

About the Opportunity:
CyberAgent AI Lab is seeking research scientists and applied research scientists to join our team.
We conduct advanced theoretical research in economics, the fusion of economics and machine learning. In addition, we conduct applied research using data from inside and outside of the company.

In this position, you will conduct theoretical and applied research in economics and related fields. You will be expected to contribute to the top international conferences and journals such as Econometrica.　

We are actively contributing to top conferences in artificial intelligence and machine learning, such as AAAI, ICML, and NeurIPS, and also are steadily producing results.
In addition, we are actively collaborating with academic economics researchers in Japan and overseas. We are also supporting the Japan Economic Association.

Press release on our collaborative research and related papers
- ​​https://www.cyberagent.co.jp/news/detail/id=22571　
　Paper: https://arxiv.org/abs/1809.03084（AAAI2019）
- https://www.cyberagent.co.jp/news/detail/id=22593
　Paper: https://arxiv.org/abs/1908.08936　（SUM2020）
- https://www.cyberagent.co.jp/news/detail/id=23320
- https://www.cyberagent.co.jp/news/detail/id=25930

Research theme examples
-Prediction of Causal Effect(CATE/ITE/Uplift Modeling)
-Application of Mechanism Design
-Demand Estimation
-Pricing
-Evaluation of Bandit / Reinforcement Learning Algorithm
-Data Fusion
※The theme is not limited to the above


Job Description
1. Research Scientist
We conduct economics and machine learning related research to be published at international conferences /journals. In addition, team management, recruitment activities, and public relations will be required depending on your position.
(However, non-research tasks can be limited upon request.)

2. Applied Research Scientist
Our mission is to apply economics to products and to contribute towards social implementation of economics in collaboration with academia and governments.
We are also developing Digital Transformation business such as in retail marketing or medical fields and we aim to provide implementations and algorithms in a wide range of areas near future. By utilizing and analyzing real data, it works both effectively and efficiently in practice to provide appropriate solutions.

-Please check below for more details (in Japanese).
https://cyberagent.ai/ailab/research/econ/
https://www.cyberagent.co.jp/way/features/list/detail/id=24538?_ga=2.83083798.1663726667.1626683080-1778396278.1596701950
https://www.cyberagent.co.jp/news/detail/id=27355
https://www.cyberagent.co.jp/news/detail/id=25930

- Publications on the top conferences
- Efficient Counterfactual Learning from Bandit Feedback（AAAI2019）
　https://arxiv.org/abs/1809.03084
- Efficient Adaptive Experimental Design for Average Treatment Effect Estimation (NeurIPS2019)
　https://arxiv.org/abs/2002.05308
- A Feedback Shift Correction in Predicting Conversion Rates under Delayed Feedback (WWW2020)
　https://arxiv.org/abs/2002.02068
- Counterfactual Cross-Validation: Stable Model Selection Procedure for Causal Inference Models (ICML2020)
　https://arxiv.org/abs/1909.05299
・Online-to-Offline Advertisements as Field Experiments（The Japanese Economic Review）
　https://link.springer.com/article/10.1007/s42973-021-00101-y
・Smartphone- and Smartwatch-acquired Daily Steps, Activity, and Barometric Pressures Associated with Subjective Measures of Rheumatoid Arthritis: a Prospective Study for RA Digital Phenotyping（eular2021）
　https://ard.bmj.com/content/80/Suppl_1/1099.3
・Non-negative Bregman Divergence Minimization for Deep Direct Density Ratio Estimation（ICML2021）
　https://arxiv.org/abs/2006.06979
・Debiased Off-Policy Evaluation for Recommendation Systems（RecSys 2021）
　https://arxiv.org/abs/2002.08536
・Adaptive Doubly Robust Estimator（NeurIPS 2021、ICML 2021 Workshop）
　https://arxiv.org/abs/2010.03792
・Learning Causal Relationship from Conditional Moment Condition by Importance Weighting（NeurIPS workshop on Deep Reinforcement Learning2021）
　https://arxiv.org/abs/2108.01312
・Mean Variance Efficient Reinforcement Learning（NeurIPS workshop on Deep Reinforcement Learning2021）
https://arxiv.org/pdf/2010.01404.pdf
・A Real-World Implementation of Unbiased Lift-based Bidding System（IEEE Big Data 2021）
　https://arxiv.org/abs/2202.13868
・Off-Policy Exploitability-Evaluation in Two-Player Zero-Sum Markov Games（AAMAS2021）
　https://arxiv.org/abs/2105.09579
・Thresholded Lasso Bandit（ICML2022）
　https://arxiv.org/abs/2010.11994
・Learning Causal Models from Conditional Moment Restrictions by Importance Weighting（ICLR2022）
　https://openreview.net/forum?id=7twQI5VnC8



Minimum Qualification Requirements for research scientists and applied scientists
- Native or business-level proficiency in Japanese
- Ph.D. in Economics or a closely related field.
- Ph.D. in Computer Science who is interested in economics is also welcomed.
- Ability to write papers as a first author and to be published at peer-reviewed conferences and journals.
- Ability to Replicate research results and conduct follow-up research on top of existing papers.
- Proficient skills in R or/and Python, and other statistical analyzing tools.
- Ability to collaborate with members from various backgrounds, e.g., researchers from the other disciplines, data scientists, and engineers

Preferred Qualification Requirements for research scientists
- Knowledge of computer science
- Work or research experience with statistics

Applied Research Scientist is also required to have applied scientists
- Strong interest in social implementation based on economics research
- Native or business-level proficiency in Japanese

Application Requirements:
CV/Resume
Research papers/link to the papers