Publications

Journals

2025

1. Masanobu Kii, Atsunori Ichikawa, Takayuki Miura. "Lightweight Two-Party Secure Sampling Protocol for Differential Privacy." Proceedings on Privacy Enhancing Technologies 2025 (PoPETs2025), Issue 1, pages 23–36, doi:10.56553/popets-2025-0003. [HP]

2. Tomoya Matsumoto, Takayuki Miura, Shingo Okamura, Behrouz Zolfaghari, Naoto Yanai. "Quo Vadis Diffusion Model? A Case for Membership Inference Attacks." IEICE Transactions on Information and Systems (2025). (To appear)

3. Osamu Saisho, Takayuki Miura, Kazuki Iwahana, Masanobu Kii, Rina Okada. "Active Synthetic Data Generation with Joint Consideration of Differential Privacy and Labeling Efficiency."  Journal of Information Processing (2025). (Recommended Paper). (To appear)

2024

1. Mine Arai, Koki Tejima, Yuya Yamada, Takayuki Miura, Kyosuke Yamashita, Chihiro Kado, Rei Shimizu, Masataka Tatsumi, Naoto Yanai, Goichiro Hanaoka. "REN-A.I.: A Video Game for AI Security Education Leveraging Episodic Memory." IEEE Access, vol.12, pp. 47359-47372, doi:10.1109/ACCESS.2024.3377699.

2. Takayuki Miura, Masanobu Kii, Toshiki Shibahara, Kazuki Iwahana, Tetsuya Okuda, Atsunori Ichikawa, Naoto Yanai. "SETSUBUN: Revisiting Membership Inference Game for Evaluating Synthetic Data Generation." Journal of Information Processing (2024). vol.32, pages 757-766, doi:10.2197/ipsjjip.32.757. [HP]

2023

1. Takayuki Miura, Toshiki Shibahara, Masanobu Kii, Atsunori Ichikawa, Juko Yamamoto, Koji Chida. "On Rényi Differential Privacy in Statistics-Based Synthetic Data Generation." Journal of Information Processing (2023). vol.31, pages 812-820, doi:10.2197/ipsjjip.31.812. (Recommended Paper). [HP]

Conference Proceeding Papers  (Refereed)

2025

1. Osamu Saisho, Takayuki Miura, Kazuki Iwahana, Masanobu Kii, Rina Okada. "Efficient Privacy-Preserving Data Annotation via Active PrivBayes Synthetic Data Generation." in 2025 IEEE International Conference on Pervasive Computing and Communications (PerCom 2025). Washington DC. March 2025. (WiP Session) [HP]

2. Saki Hiromi, Hiroki Kinoshita, Masanori Yamada, Takayuki Miura. "Enhancing Jailbreak Resistance in Large Language Models Using Model Merge." in 8th Deep Learning Security and Privacy Workshop (DLSP 2025). San Francisco, CA. May 2025. [HP]

3. Toshiki Shibahara, Takayuki Miura, Masanobu Kii, Atsunori Ichikawa. "Membership Inference Attack Against Bayesian Neural Network." in ICC 2025-IEEE International Conference on Communications (ICC 2025). Montreal, Canada. June 2025. [HP]

4. Juko Yamamoto, Takayuki Miura, Rina Okada, Masanobu Kii, Atsunori Ichikawa. "Explaining and Visualizing Synthetic Data Quality Using Statistical Distances." in 2025 International Workshop on Privacy Engineering (IWPE'25). Venice, Italy. June 2025. [HP]

5. Yusuke Yamasaki, Kenta Niwa, Daiki Chijiwa, Takumi Fukami, Takayuki Miura. "Plausible Token Amplification for Improving Accuracy of Differentially Private In-Context Learning Based on Implicit Bayesian Inference." in  International Conference on Machine Learning (ICML2025), Vancouver, Canada. July 2025. [HP]

6. Yuki Wakai, Toshiki Shibahara, Rina Okada, Takayuki Miura, Hiroki Kinoshita. "Understanding Privacy Risks of Large Language Models in Japanese Based on Training Data Extraction Attacks." in the AsiaCCS 2025 Workshop on Privacy in Large Language Models and Natural Language Processing (LM-SHIELD2025). Hanoi, Vietnam. August 2025. [HP][Paper]

2024

1. Takayuki Miura, Eizen Kimura, Atsunori Ichikawa, Masanobu Kii, Juko Yamamoto. "Evaluating Synthetic Data Generation Techniques for Medical Dataset". in 17th International Conference on Health Informatics (HEALTHINF 2024). Rome, Italy. Feburary 2024. [Paper]

2. Toshiki Shibahara, Takayuki Miura, Masanobu Kii, Atsunori Ichikawa. "Efficiently Calculating Stronger Lower Bound for Differentially Private SGD in Black-box Setting." in 2024 IEEE 48th Annual Computers, Software, and Applications Conference (COMPSAC 2024). Osaka, Japan. July 2024.

3. Takayuki Miura, Toshiki Shibahara, Naoto Yanai. "MEGEX: Data-Free Model Extraction Attack Against Gradient-Based Explainable AI." in The 2nd ACM Workshop on Secure and Trustworthy Deep Learning Systems (SecTL 2024). pages 56-66. Singapore. July 2024. [HP][Paper]

4. Iifan Tyou, Shigenori Ohashi, Justin Yu, Takayuki Miura, Takuro Hosoi, Kanta Matsuura. "Leveraging Timestamps to Create Secure and Feeless Evidence Management." in the 7th IEEE International Conference on Blockchain (Blockchain 2024). Copenhagen, Denmark. August 2024. [HP] [Paper]

2023

1. Tomoya Matsumoto, Takayuki Miura, Naoto Yanai. "Membership Inference Attacks against Diffusion Models." in 6th Deep Learning Security and Privacy Workshop (DLSP 2023). San Francisco, CA. May 2023. [preprint] [HP]

2. Kazuki Iwahana, Osamu Saisho, Takayuki Miura, Akira Ito. "Can Inputs' Reconstruction Information be Coded into Machine Learning Model's Outputs?" in SECAI: Workshop on Security and Artificial Intelligence (SECAI 2023). Hague, Netherlands. September 2023. [HP]

2020

1. Shinsuke Ito, Takayuki Miura, Hiroto Akatsuka, Masayuki Terada. "Differential privacy and its applicability for official statistics in Japan–a comparative study using small area data from the Japanese population census." in Privacy in Statistical Databases: UNESCO Chair in Data Privacy, International Conference (PSD 2020), Tarragona, Spain, September 23–25, 2020, Proceedings, pp. 337-352. Springer International Publishing, 2020. [HP]

Conference Papers  (Refereed but No Proceedings)

2025

1. Tomoya  Matsumoto, Takayuki Miura, Toshiki Shibahara, Masanobu Kii, Kazuki Iwahana, Osamu Saisho, Shingo Okamura. "Differentially Private Sequential Data Synthesis with Structured State Space Model and Diffusion Model." in the Theory and Practice of Differential Privacy (TPDP2025). Mountain View, CA. June 2025. [HP]

2024

1. Osamu Saisho, Takayuki Miura, Kazuki Iwahana, Masanobu Kii, Rina Okada. "Active Learning for Human Annotation of Privacy-Preserved Synthetic Data." in Privacy in Statistical Databases: UNESCO Chair in Data Privacy, International Conference (PSD 2024), Antibes Juan-les-Pins, France. September 25–27, 2024. [HP]

2. Tomoya  Matsumoto, Takayuki Miura, Toshiki Shibahara, Masanobu Kii, Kazuki Iwahana, Osamu Saisho, Shingo Okamura. "Differentially Private Sequential Data Synthesis with Structured State Space Models and Diffusion Models." in the NeurIPS Safe Generative AI Workshop. Vancouver, Canada. December 2024. [HP]

3. Tomoya Yamashita, Takayuki Miura, Yuki Yamanaka, Toshiki Shibahara, Masanori Yamada. "Concept Unlearning for Large Language Models.", in the NeurIPS Safe Generative AI Workshop. Vancouver, Canada. December 2024. [HP]

Talks

2023

1. The 18th International Workshop on Security (IWSEC2023), SCIS/CSS Session. "On Rényi Differential Privacy in Statistics-Based Synthetic Data Generation." Yokohama, Japan. August 31, 2023. (Invited Talk). [HP]

Patents

1. US 12242627 B2, Retention-replacement probability generation device, retention-replacement perturbation device, retention-replacement probability generation method, retention-replacement perturbation method, and program, Takayuki Miura, Satoshi Hasegawa, 2025-03-04. [Link]

2. US 12124612 B2, Identification estimate risk evaluation apparatus, identification estimate risk evaluation method, and program, Satoshi Hasegawa, Takayuki Miura, 2024-10-22. [Link]

Preprints (arXiv, ePrint)

2025

1. Tomoya Yamashita, Yuuki Yamanaka, Masanori Yamada, Takayuki Miura, Toshiki Shibahara, Tomoharu Iwata. "Concept Unlearning in Large Language Models via Self-Constructed Knowledge Triplets." arXiv preprint arXiv:2509.15621 (2025). [Link]

2. Tomoya Yamashita, Akira Ito, Yuuki Yamanaka, Masanori Yamada, Takayuki Miura, Toshiki Shibahara. "Sparse-Autoencoder-Guided Internal Representation Unlearning for Large Language Models." arXiv preprint arXiv:2509.15631 (2025). [Link]

3. Akira Ito, Takayuki Miura, Yosuke Todo. "Is the Hard-Label Cryptanalytic Model Extraction Really Polynomial?" arXiv preprint arXiv:2510.06692 (2025). [Link] [ePrint]

4. Kazuki Iwahana, Yusuke Yamasaki, Akira Ito,  Takayuki Miura, Toshiki Shibahara. "Robust Backdoor Removal by Reconstructing Trigger-Activated Changes in Latent Representation." arXiv preprint arXiv:2511.08944 (2025). [Link]

2023

1. Takayuki Miura, Toshiki Shibahara, Masanobu Kii, Atsunori Ichikawa, Juko Yamamoto, Koji Chida. "On Rényi Differential Privacy in Statistics-Based Synthetic Data Generation." arXiv preprint arXiv:2303.17849 (2023).

2. Tomoya Matsumoto, Takayuki Miura, Naoto Yanai. "Membership inference attacks against diffusion models." arXiv preprint arXiv:2302.03262 (2023).

2021

1. Takayuki Miura, Satoshi Hasegawa, Toshiki Shibahara. "MEGEX: Data-free model extraction attack against gradient-based explainable AI." arXiv preprint arXiv:2107.08909 (2021).

2019

1. Takayuki Miura. "Classification of del Pezzo surfaces with 1/3(1, 1)- and 1/4(1, 1)-singularities." arXiv preprint arXiv:1903.00679 (2019).