Rakuten Institute of Technology

Congratulations to Sehyung Lee, Yeongnam Chae, Mijung Kim, and Bjorn Stenger from Rakuten Institute of Technology on their paper's acceptance at ECCV 2024! Their work, "Linearly Controllable GAN: Unsupervised Feature Categorization and Decomposition for Image Generation and Manipulation," offers innovative advancements in realistic facial image generation with fine-grained control. ECCV 2024 will take place from September 29 to October 4 in Milano, Italy. Well done, team! https://eccv.ecva.net/

Congratulations to Chandrai Kayal, Sayantan Chattopadhyay, Aryan Gupta, Satyen Abrol, and Archie Gugol, from the Rakuten Institute of Technology – India! Their paper, “JLBert: Japanese Light BERT for Cross-Domain Short Text Classification,” has been accepted at LREC-COLING 2024 in the regular short track! acceptance at LREC-COLING 2024 in the regular short track! This novel model is designed to enhance Short Text Classification (STC) tasks in Japanese, showcasing our commitment to cutting-edge research and development. Chandrai represented the team at the Joint International Conference on Computational Linguistics, Language Resources and Evaluation, which was held from May 20 to May 25 in Torino, Italy. Kudos to the team on this fantastic work!

Congratulations to Md Mostafizur Rahman, Daisuke Kikuta (Ex-intern, RIT), Yu Hirate, Researchers at Rakuten Institute of Technology in Tokyo, along with Toyotaro Suzumura, collaborator from the University of Tokyo, on the acceptance of their paper at the 47th International ACM SIGIR Conference on Research and Development in Information Retrieval for presentation. Their paper, titled "Graph-Based Audience Expansion Model for Marketing Campaigns," introduces AudienceLinkNet, a novel solution for audience expansion in the context of Rakuten's diverse services and clients. AudienceLinkNet tackles challenges such as a heterogeneous user base, complex marketing campaigns, and limited seed users by formulating the audience expansion problem as a graph problem and combining the Pre-trained Knowledge Graph Embedding Model with Graph Convolutional Networks.   Don't miss the chance to catch their presentation at the 47th International ACM SIGIR Conference (https://lnkd.in/gT92KiU8), happening on July 14-18, 2024, in Washington D.C., USA. #SIGIR2024 #InformationRetrieval #AudienceExpansion #RIT #Rakuten

Dr. Tracy Holloway King delivered an insightful talk on "Multi-modal Search: To embed or not to embed" at Rakuten Group Technology Services Division Tech Talk on April 10th. She highlighted the challenges users face in multi-modal and cross-modal search and shared game-changing techniques for image, video, and audio search employed by Adobe. The session saw huge participation and sparked lively discussions, making it an engaging and informative event!

Our paper, titled “Search Query Refinement for Japanese Named Entity Recognition in E-commerce Domain”, by Yuki Nakayama (中山 祐輝), Ryutaro Tatsushima, Erick Mendieta, Koji Murakami and Keiji Shinzato from Rakuten Institute of Technology, has been accepted for the NAACL 2024 Industry Track! This paper presents an innovative approach that significantly improves query refinement by 3.5 points compared to the baseline. The research also shows an impressive 11.7-point performance boost on E-commerce specific Named Entity Recognition tasks. The conference (https://2024.naacl.org/) will take place from June 16-21 in Mexico City, Mexico. #JapaneseNLP #QueryRefinement #NAACL2024 #NamedEntityRecognition

Congratulations to Debopriyo Banerjee and Ashish Kulkarni, Research Scientists at Rakuten Institute of Technology in India, along with Debrup Das and Somak Aditya, collaborators from IIT Kharagpur, on acceptance of their long paper at NAACL 2024, and poster paper at ICLR 2024 Workshop ME-FoMo.   Their long paper, titled "MATHSENSEI: A Tool-Augmented Large Language Model for Mathematical Reasoning" explores the benefits of using Tool-Augmented large Language Models (TALMs) in the context of mathematical reasoning, which achieved impressive results, outperforming vanilla Chain-of-Thought (CoT) prompting method by 13.5% on the MATH dataset. Exciting advancements in Mathematical Reasoning!    Pre-print paper: https://lnkd.in/gqNSXMT3 Official open-source code: https://lnkd.in/gJ4AxX5x Project Page: https://lnkd.in/gxi-aUUW   This year, the NAACL conference (https://2024.naacl.org/) will take place from June 16-21 in Mexico City, Mexico, and the ICLR 2024 Workshop ME-FoMo (https://lnkd.in/gJMURTEm) on May 11, 2024 in Vienna, Austria.   #NAACL2024 #ComputationalLinguistics #MathematicalReasoning #ICLR2024 #MEFoMo2024

Exciting news! Our short paper, titled "Position Bias Estimation with Item Embedding for Sparse Dataset" by Shion Ishikawa, Yun-Ching Liu, Young-joo Chung and Yu Hirate from Rakuten Institute of Technology, has been accepted for the 2024 ACM Web Conference. Our research addresses the challenge of estimating position bias in Learning to Rank systems. We propose a variant of the REM algorithm that uses item embeddings to handle sparsity in real-world scenarios.   The conference will be held from 13-17 May 2024 in Singapore. https://lnkd.in/ga8wBMex   Stay tuned for more updates on our findings! #ACMconference #learningtorank #EMalgorithm #sparsity

Congratulations to Chandrai Kayal, Sayantan Chattopadhyay, Aryan Gupta, Satyen Abrol and Archie Gugol, members of Rakuten Institute of Technology, for their paper acceptance at LREC-COLING 2024 in the regular short track! Chandrai’s paper, titled “JLBert: Japanese Light BERT for Cross-Domain Short Text Classification”, introduces a novel cross-domain model called JLBert, specifically designed to enhance Short Text Classification (STC) tasks in Japanese. The 2024 Joint International Conference on Computational Linguistics, Language Resources and Evaluation will be held on May 20-25 in Torino, Italy. https://lnkd.in/gvYjN_Ut

We are thrilled to announce that a research paper authored by Rakuten Institute of Technology (RIT) in Boston has been accepted for presentation at the IEEE BigData 2023 conference! This collaborative effort, involving our former intern Tianqi Wang, has paved the way for RIT to explore innovative approaches in addressing label noise during supervised learning. Title: A Prior-Guided Meta-Weight Network for Long-tailed Noisy Data in Item Categorization Authors: Tianqi Wang (Former Intern), Lei Chen (RIT), and Jing Gao (Associate professor at Purdue University) https://lnkd.in/gy87VDmk

Rakuten Medical is excited to share that we presented two posters of AI-based study on immune characteristics in responders and cellular level drug quantification of #Alluminox™ treatment (#photoimmunotherapy) at #SITC2023. Learn more at: https://lnkd.in/e__MKhzw