Skip to content
master
Switch branches/tags
Code

Latest commit

 

Git stats

Files

Permalink
Failed to load latest commit information.
Type
Name
Latest commit message
Commit time
 
 
 
 

Yohei KIKUTA, PhD
Tokyo, Japan
diracdiego[at]gmail.com

Summary

  • Expert at machine learning, 5+ years experience in both research and practical application
  • Solved machine learning/data analysis problems for business purposes: image recognition, recommendation, natural language processing, and so forth
  • Published academic papers about deep learning and recommendation
  • Made many external presentations about machine learning
  • Contributed article to NikkeiBP and JSAI, and wrote book for beginners of machine learning
  • Core competencies: Machine learning theory and implementation, image analysis, recommendation, Python, R

Skills

Machine Learning

  • Deep theoretical understanding of machine learning
  • Image recognition, recommender system, natural language processing, marketing ROI optimization, time series analysis
  • TensorFlow, PyTorch, scikit-learn, gensim, and so many more

DevOps

  • Basic knowledge of Linux (Ubuntu) and MINIX
  • Git (GitHub), Docker (Docker Hub) and CI (CircleCI and Jenkins)
  • AWS, GCP

Language

  • Japanese (native), English (business)
  • Python (expert), C (advanced), C++ (intermediate), Kotlin (beginner), Haskell (begineer)

Experience

Ubie, Inc (202004 - Present)

  • Machine learning engineer & Data scientist

Self-employement (201902 - 202003)

  • Entrusted project from a laboratory
    • Quality assurance of machine learning applications
      Surveyed and summarized recent progress on adversarial examples (+50 papers), code implementations using PyTorch.
  • Self-improvement
    • Theoretical topics
      Read various machine learning papers, studied information/measure theory.
    • Code constructions
      Reimplementation of machine learning papers, made simple Android apps, competitive programming, etc.
    • Basics of computer science
      CPU (ARM and X86), OS such as processes/threads and memory management (Linux and MINIX), interpreter/compiler (implementation of a PostScript interpreter in C), etc.

Cookpad Inc. (201612 - 201901)

  • Research and Development
    • Created new service using machine learning
      Image recognition using deep learning: classification, object detection, attractiveness estimation, etc
    • Attended machine learning conference
      PAKDD, IJCAI, NIPS (as a top conference reporter of JSAI in 2017), ...
  • Sponsoring conference and organizing event
    • Led company's exhibition, chaired a session of IPSJ, is one of the committee members of CEA
    • Organized various events, see connpass page
  • Hiring and team building
    • Conducted interview with candidates for machine learning position in domestic and global
    • Contributed team building and communication strategy

Deloitte (201404 - 201611)

  • Research and Development
    • Investigated business application through cutting edge technologies and wrote papers 
      Image recognition using deep learning, recommendation
  • Data analysis service to a client (onsite project at a client's office)
    • Built machine learning module for a recommender system
      Learning to rank, Xgboost, Factorization Machines, Spark, Hive, AWS
    • Analyzed various topics: ROI optimization, time series analysis for the budget planning, etc
      ARIMA, price elasticity analysis, linear programming, genetic algorithm
  • Characteristic analysis of tourist spots and brand reputation analysis of a company
    • Analyzed for reporting to clients
      Text data processing, topic model (PLSA), bayesian network
  • Joint research about business applications using deep learning

Publications

Papers

  • SRGAN for Super-Resolving Low-Resolution Food Images (conference link)
    • IJCAI-ECAI2018 WS CEA2018, poster, 20180715
  • Improving SRGAN for Super-Resolving Low Resolution Food Images (link)
    • JSAI2018, in Japanese, 20180607
  • ClassSim: Similarity between Classes Defined by Misclassification Ratios of Trained Classifiers (link)
  • Approaches to Food/Non-food image classification using Deep Learning on cookpad (link)
    • IJCAI2017 WS39 CEA2017, poster, 20170820
  • Cookpad Image Dataset: An Image Collection as Infrastructure for Food Research (link)
    • SIGIR2017, resource paper, 20170807
  • Web-Scale Personalized Real-Time Recommender System on Suumo (conference link)
    • PAKDD2017, long paper, 20170525
  • Approaches to Food/Non-food image classification using Deep Learning on cookpad (link)
    • JSAI2017, in Japanese, 20170523
  • Proposing automated region extraction techniques from image data (link)
    • JSAI2016, in Japanese, 20160606
  • Inappropriate image detection based on Deep Learning (link)
    • JSAI2016, in Japanese, 20160606
  • Exploiting the Hidden Layer Information Toward the Understanding and Utilization of Feature Representations Obtained from Deep Learning (link)
    • JSAI2015, in Japanese, 20150531
  • Physics papers during my Ph.D. student years (link)

Books & Articles, Blogs, Interviews

  • Books & Articles
    • A Primer on Adversarial Examples (link)
      • 20200227, 技術書展8, in Japanese
    • フリーライブラリで学ぶ機械学習入門 (link)
      • 20170321, 秀和システム, in Japanse
      • Contribution to chapters 1,6 and 8.
    • Web articles
      • 20180515, 機械学習を用いた画像分類における「未解決問題」を解くためにやったこと(link), in Japanese
      • 20180620, GeekOutナイト(link1, link2, link3, link4), in Japanese
    • 人工知能学会誌寄稿
      • 201805, 会議報告「The Thirty-first Annual Conference on Neural Information Processing Systems(NIPS 2017)」 (link), in Japanese
      • 201809, 「AI トレンド・トップカンファレンス NIPS 2017」報告会 (link), in Japanese
    • この1冊でまるごとわかる! 人工知能ビジネス (link)
      • 20150829, 日経BP, in Japanse
      • Contribution to the article of p.86-87.
  • Blogs
    • 原理的には可能 (In Japanese)
    • COOKPAD Engineer's Blog (in Japanese)
      • 20181204, BERT with SentencePiece で日本語専用の pre-trained モデルを学習し、それを基にタスクを解く (link)
      • 20180831, Cookpad Summer Internship 2018 5 DAY R&D を開催しました (link)
      • 20180705, Firebase ML Kitで自作のカスタムモデルを使って料理・非料理画像を判定できるようにした (link)
      • 20180328, 人工知能学会のトップカンファレンス派遣レポータとして NIPS2017 に参加しました (link)
      • 20170914, 料理きろくにおける料理/非料理判別モデルの詳細 (link)
      • 20170809, 2nd Hackarade: Machine Learning Challenge (link)
  • Interviews
    • 「機械学習で食生活を豊かにする」ことに挑む物理学博士が思い描く研究とサービスの良い関係 (前編, 後編)
      • 20180807, 20180810, forkwell press, in Japanese
    • クックパッドにおける料理きろくサービスと研究開発 (link)  
      • 20171223, 日刊工業新聞, in Japanese

Presentations

  • Full list
  • Speaker Deck
  • Many presentations in private study groups
    • Pattern Recognition and Machine Learning, Deep Learning, Python Machine Learning, Deep Learning with python, Information Theory Inference and Learning Algorithms, Categories types and structures, 深層学習, はじめてのパターン認識, 詳解ディープラーニング, 経済・ファイナンスデータの計量時系列分析, データ解析のための統計モデリング入門, etc

Podcast

Education

  • Graduate University for Advanced Studies (200904 - 201403)
    • Doctor of Philosophy (Ph.D.) of Elementary Particle Physics
    • Ph.D. thesis: Higgs interactions in physics beyond the standard model
    • JSPS Research Fellowship for Young Scientists (DC2)
  • Tohoku University (200504 - 200903)
    • Bachelor's degree in Physics

Ohters

About

my resume

Resources

License

Releases

No releases published

Packages

No packages published