Takashi Nishide
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Takashi Nishide has a total of 52 co-authors. Here are the top co-authors by number of shared publications:
- Eiji Okamoto
University of Tsukuba
3 shared publications - Goichiro Hanaoka
National Institute of Advanced Industrial Science and Technology, Tokyo, Japan
3 shared publications - Hikaru Tsuchida
Saitama Institute of Technology
3 shared publications - Hiroki Okada
KDDI Research, Inc
3 shared publications - Kazuhide Fukushima
KDDI Research, Inc
3 shared publications - Shintaro Narisada
KDDI Research, Inc
3 shared publications
Latest Publications
- 19 October 2026
- Published by Springer Nature
- p. 530-551
Multi-party Computation Over Finite Cyclic Groups with Friends and Foes
- 19 October 2026
- Published by Springer Nature
- p. 530-551
Multi-party Computation Over Finite Cyclic Groups with Friends and Foes
- 18 October 2026
- Published by Springer Nature
- p. 206-224
Fast MPC Framework Under Shamir’s Secret Sharing with Malicious Security and Two-Thirds Honest Majority
- 18 October 2026
- Published by Springer Nature
- p. 206-224
Fast MPC Framework Under Shamir’s Secret Sharing with Malicious Security and Two-Thirds Honest Majority
- Open Access
- 21 August 2026
- Published by Scilight Press Pty Ltd in Pragmatic Cybersecurity
- Vol. 1 (2) , 13
Efficient Homomorphic String Search via TFHE †
We present a method for secure pattern matching over encrypted texts using TFHE. Our approach realizes a fully secure binary search algorithm by leveraging two operational modes of integer-input TFHE.While the BGV-based method of Bonte and Iliashenko (CCSW’20) requires O(|P|·|T|) secure character comparisons to find a pattern P in a text T, our method reduces this to O(|P| log |T|) comparisons, achieving improved scalability for large texts. As a result, our method can find a pattern of length 100 in an encrypted text containing genomic data of one million characters in less than 5 min, where prior work would require approximately 5 days for the same task. These results highlight the practicality of TFHE and its potential for large-scale secure string search.
- 21 August 2026
- Published by Scilight Press Pty Ltd in Pragmatic Cybersecurity
- Vol. 1 (2) , 13
Open AccessEfficient Homomorphic String Search via TFHE †
We present a method for secure pattern matching over encrypted texts using TFHE. Our approach realizes a fully secure binary search algorithm by leveraging two operational modes of integer-input TFHE.While the BGV-based method of Bonte and Iliashenko (CCSW’20) requires O(|P|·|T|) secure character comparisons to find a pattern P in a text T, our method reduces this to O(|P| log |T|) comparisons, achieving improved scalability for large texts. As a result, our method can find a pattern of length 100 in an encrypted text containing genomic data of one million characters in less than 5 min, where prior work would require approximately 5 days for the same task. These results highlight the practicality of TFHE and its potential for large-scale secure string search.
- 1 May 2026
- Published by Springer Nature
- p. 91-106
Publicly Verifiable Multi-server Homomorphic Secret Sharing
- 1 May 2026
- Published by Springer Nature
- p. 91-106
Publicly Verifiable Multi-server Homomorphic Secret Sharing
- 11 December 2025
- Published by Springer Nature
- p. 365-389
One-Time Memories Secure Against Depth-Bounded Quantum Circuits
- 11 December 2025
- Published by Springer Nature
- p. 365-389
One-Time Memories Secure Against Depth-Bounded Quantum Circuits
- Open Access
- 7 July 2025
- Published by International Association for Cryptologic Research in IACR Communications in Cryptology
- Vol. 2 (2)
Lattice-based Multi-key Homomorphic Signatures Forward-unforgeable against Signing Key Leakage
Homomorphic signature (HS) schemes enable an untrusted server to run some computation over the data signed under the same key and derive a short signature for authenticating the computation result. Fiore et al. (Asiacrypt'16) introduced novel lattice-based multi-key homomorphic signatures (MKHS) to support an evaluation of signatures under multiple/different keys, and anyone can verify the resultant signature by using corresponding public verification keys. However, a limitation of their scheme is that even if only one signing key is leaked, a malicious server can forge a signature on a fake computation result involving the inputs of uncorrupted signers. To address this issue, we propose a new scheme built upon the work of Fiore et al., aiming to achieve a stronger security guarantee, which we call forward unforgeability, against signing key leakage. Our MKHS scheme is constructed based on the short integer solution (SIS) problem as Fiore et al., and can be forward-unforgeable even if an adversary obtains all the signing keys. Furthermore, we propose a variant by introducing a helper entity to amortize the overhead of signature verifications.
- 7 July 2025
- Published by International Association for Cryptologic Research in IACR Communications in Cryptology
- Vol. 2 (2)
Open AccessLattice-based Multi-key Homomorphic Signatures Forward-unforgeable against Signing Key Leakage
Homomorphic signature (HS) schemes enable an untrusted server to run some computation over the data signed under the same key and derive a short signature for authenticating the computation result. Fiore et al. (Asiacrypt'16) introduced novel lattice-based multi-key homomorphic signatures (MKHS) to support an evaluation of signatures under multiple/different keys, and anyone can verify the resultant signature by using corresponding public verification keys. However, a limitation of their scheme is that even if only one signing key is leaked, a malicious server can forge a signature on a fake computation result involving the inputs of uncorrupted signers. To address this issue, we propose a new scheme built upon the work of Fiore et al., aiming to achieve a stronger security guarantee, which we call forward unforgeability, against signing key leakage. Our MKHS scheme is constructed based on the short integer solution (SIS) problem as Fiore et al., and can be forward-unforgeable even if an adversary obtains all the signing keys. Furthermore, we propose a variant by introducing a helper entity to amortize the overhead of signature verifications.
- 1 February 2025
- Published by Springer Nature
- p. 43-62
Secure Five-Party Computation with Private Robustness and Minimal Online Communication
- 1 February 2025
- Published by Springer Nature
- p. 43-62
Secure Five-Party Computation with Private Robustness and Minimal Online Communication
- Open Access
- 1 January 2025
- Published by Information Processing Society of Japan in Journal of Information Processing
- Vol. 33, 583-593
VeHSS: Two-server Verifiable Homomorphic Secret Sharing Leveraging SIMD Operations
Outsourced computation poses challenges in data privacy and computation integrity. Fully homomorphic encryption (FHE) ensures data privacy but incurs high computational overhead. Homomorphic secret sharing (HSS), an alternative with reduced overhead, enables homomorphic evaluations to be distributed across remote servers without interaction. On the downside, neither FHE nor HSS guarantees computation integrity. To address this issue, verifiable homomorphic secret sharing (VHSS) schemes have been proposed to verify computation correctness. However, existing VHSS schemes for polynomials only verify whether the servers perform the same function rather than the intended specified function, and implicitly assume that at least one server is honest. Moreover, the costs of generating verification information are the same as or even more than re-executing the computation. We propose a two-server VHSS scheme leveraging single-instruction multiple data (SIMD) parallel computations. Our scheme verifies computation correctness for specified functions even when both non-colluding servers are malicious under our security model. Moreover, it supports amortized verification on the client side by enabling the precomputation of reusable values for verification, while introducing no additional computational costs on servers. As a byproduct, we also discuss how our method can be applied to FHE to mitigate recent attacks targeting decryption keys.
- 1 January 2025
- Published by Information Processing Society of Japan in Journal of Information Processing
- Vol. 33, 583-593
Open AccessVeHSS: Two-server Verifiable Homomorphic Secret Sharing Leveraging SIMD Operations
Outsourced computation poses challenges in data privacy and computation integrity. Fully homomorphic encryption (FHE) ensures data privacy but incurs high computational overhead. Homomorphic secret sharing (HSS), an alternative with reduced overhead, enables homomorphic evaluations to be distributed across remote servers without interaction. On the downside, neither FHE nor HSS guarantees computation integrity. To address this issue, verifiable homomorphic secret sharing (VHSS) schemes have been proposed to verify computation correctness. However, existing VHSS schemes for polynomials only verify whether the servers perform the same function rather than the intended specified function, and implicitly assume that at least one server is honest. Moreover, the costs of generating verification information are the same as or even more than re-executing the computation. We propose a two-server VHSS scheme leveraging single-instruction multiple data (SIMD) parallel computations. Our scheme verifies computation correctness for specified functions even when both non-colluding servers are malicious under our security model. Moreover, it supports amortized verification on the client side by enabling the precomputation of reusable values for verification, while introducing no additional computational costs on servers. As a byproduct, we also discuss how our method can be applied to FHE to mitigate recent attacks targeting decryption keys.
- Open Access
Local Curvature Smoothing with Stein's Identity for Efficient Score Matching
The training of score-based diffusion models (SDMs) is based on score matching. The challenge of score matching is that it includes a computationally expensive Jacobian trace. While several methods have been proposed to avoid this computation, each has drawbacks, such as instability during training and approximating the learning as learning a denoising vector field rather than a true score. We propose a novel score matching variant, local curvature smoothing with Stein's identity (LCSS). The LCSS bypasses the Jacobian trace by applying Stein's identity, enabling regularization effectiveness and efficient computation. We show that LCSS surpasses existing methods in sample generation performance and matches the performance of denoising score matching, widely adopted by most SDMs, in evaluations such as FID, Inception score, and bits per dimension. Furthermore, we show that LCSS enables realistic image generation even at a high resolution of $1024 \times 1024$.Open AccessLocal Curvature Smoothing with Stein's Identity for Efficient Score Matching
The training of score-based diffusion models (SDMs) is based on score matching. The challenge of score matching is that it includes a computationally expensive Jacobian trace. While several methods have been proposed to avoid this computation, each has drawbacks, such as instability during training and approximating the learning as learning a denoising vector field rather than a true score. We propose a novel score matching variant, local curvature smoothing with Stein's identity (LCSS). The LCSS bypasses the Jacobian trace by applying Stein's identity, enabling regularization effectiveness and efficient computation. We show that LCSS surpasses existing methods in sample generation performance and matches the performance of denoising score matching, widely adopted by most SDMs, in evaluations such as FID, Inception score, and bits per dimension. Furthermore, we show that LCSS enables realistic image generation even at a high resolution of $1024 \times 1024$. - Open Access
- 24 March 2024
- Published by Association for the Advancement of Artificial Intelligence (AAAI) in Proceedings of the AAAI Conference on Artificial Intelligence
- Vol. 38 (19) , 21492-21500
Understanding Likelihood of Normalizing Flow and Image Complexity through the Lens of Out-of-Distribution Detection
Out-of-distribution (OOD) detection is crucial to safety-critical machine learning applications and has been extensively studied. While recent studies have predominantly focused on classifier-based methods, research on deep generative model (DGM)-based methods have lagged relatively. This disparity may be attributed to a perplexing phenomenon: DGMs often assign higher likelihoods to unknown OOD inputs than to their known training data. This paper focuses on explaining the underlying mechanism of this phenomenon. We propose a hypothesis that less complex images concentrate in high-density regions in the latent space, resulting in a higher likelihood assignment in the Normalizing Flow (NF). We experimentally demonstrate its validity for five NF architectures, concluding that their likelihood is untrustworthy. Additionally, we show that this problem can be alleviated by treating image complexity as an independent variable. Finally, we provide evidence of the potential applicability of our hypothesis in another DGM, PixelCNN++.
- 24 March 2024
- Published by Association for the Advancement of Artificial Intelligence (AAAI) in Proceedings of the AAAI Conference on Artificial Intelligence
- Vol. 38 (19) , 21492-21500
Open AccessUnderstanding Likelihood of Normalizing Flow and Image Complexity through the Lens of Out-of-Distribution Detection
Out-of-distribution (OOD) detection is crucial to safety-critical machine learning applications and has been extensively studied. While recent studies have predominantly focused on classifier-based methods, research on deep generative model (DGM)-based methods have lagged relatively. This disparity may be attributed to a perplexing phenomenon: DGMs often assign higher likelihoods to unknown OOD inputs than to their known training data. This paper focuses on explaining the underlying mechanism of this phenomenon. We propose a hypothesis that less complex images concentrate in high-density regions in the latent space, resulting in a higher likelihood assignment in the Normalizing Flow (NF). We experimentally demonstrate its validity for five NF architectures, concluding that their likelihood is untrustworthy. Additionally, we show that this problem can be alleviated by treating image complexity as an independent variable. Finally, we provide evidence of the potential applicability of our hypothesis in another DGM, PixelCNN++.