Toru Nakamura’s research while affiliated with KDDI Research and other places

Recently, an increasing number of laws have governed the useability of users' privacy. For example, Article 17 of the General Data Protection Regulation (GDPR), the right to be forgotten, requires machine learning applications to remove a portion of data from a dataset and retrain it if the user makes such a request. Furthermore, from the security perspective, training data for machine learning models, i.e., data that may contain user privacy, should be effectively protected, including appropriate erasure. Therefore, researchers propose various privacy-preserving methods to deal with such issues as machine unlearning. This paper provides an in-depth review of the security and privacy concerns in machine learning models. First, we present how machine learning can use users' private data in daily life and the role that the GDPR plays in this problem. Then, we introduce the concept of machine unlearning by describing the security threats in machine learning models and how to protect users' privacy from being violated using machine learning platforms. As the core content of the paper, we introduce and analyze current machine unlearning approaches and several representative research results and discuss them in the context of the data lineage. Furthermore, we also discuss the future research challenges in this field.

This paper provides a solution to mitigate mistrust that arises from untransparency in a multi-stakeholder environment. This work is related to a kind of verifiable computation. When considering the multi-stakeholder environment, not only the participants’ requirements but also the assignee’s intention should be respected. That is, the assignee should be given the discretion to select a result that is the best for the assignee among all the choices. However, there is a possibility that if the assignee is malicious, he/she may falsify and provide an inadequate result for participants to maximize his/her benefit by ignoring the participants’ requirements. It is difficult for the participants to detect this if they want to keep their preference orders secret from others. This paper proposes a solution to determine whether the received result is adequate for the participants while keeping their preference orders secret. The proposed solution is based on fully homomorphic encryption (FHE) and assumes the use of a semi-honest third-party server. This paper first describes a general solution that is not limited to specific requirements from participants. Next, this paper shows a way to apply to stable matching problem as a specific implementation. More specifically, a transformation of a standard stability checking algorithm into an algorithm that can be implemented by FHE with the computational complexity O(n2)$O(n^2)$. Finally, this paper gives an example of an implementation and its performance with HElib, which is an FHE library that provides BGV.

Stable matching is an important problem that receives attention from researchers in several fields. In the problem setting, there are two sets with the same number of members. Each member has its matching preference. The goal is to find a one-to-one matching between each member of the two sets such that no pairs want to change the matching result. Since an instance of the stable matching problem may have more than one possible stable matching, Nakamura et al. proposed a multi-stakeholder environment with selectability property, and applied it to the stable matching problem as an example use case. In their setting, the computing server could freely choose to return any stable matching depending on the benefits of the clients and the computing server. Their protocol also offered verifiability, but only against a semi-honest verifying server. To address this issue, we propose a verification protocol for stable matching against a malicious server. Our verification protocol is constructed from CDS schemes for stable matching, which do not require any asymmetric-key cryptographic primitives. From the implementation result, our proposed protocol is 4 to 5 orders of magnitude faster than the previous work.

The local differential privacy metric has attracted attention due to its quantitative nature, and many mechanisms have been studied for satisfying local differential privacy based on data formats and use cases. Local differential privacy mechanisms generally target a certain data space and perturb it sufficiently to provide indistinguishability of the data on that space. Therefore, individual data tends to be greatly disturbed so that even relatively simple tasks require a large amount of data to equalize the noise caused by the mechanism. In this paper, we define hierarchical local differential privacy, which is an extension of local differential privacy, and propose a mechanism to satisfy both local differential privacy and hierarchical local differential privacy. Hierarchical local differential privacy views a data space hierarchically as a set of smaller spaces, and instead of abandoning the privacy of data contained in different spaces, the amount of noise can be reduced. In this paper, we further design a hierarchical local differential privacy framework and achieve a privacy guarantee based on local differential privacy for all the data in the framework. Finally, we experimentally evaluate the proposed framework using image data. The framework allows control over the amount of information that can be disclosed, and furthermore, maintains a higher degree of utility than applying a simple local differential privacy mechanism.

This paper delivers a post-quantum construction for group-oriented attribute-based encryption (GO-ABE) using lattice-based cryptography. The GO-ABE scheme enables users from the same group to combine their attributes to satisfy a decryption policy without revealing their secret keys. GO-ABE is particularly useful when no single user can fulfill the decryption policy alone, but a group of users can satisfy it together. Li et al. introduced the idea of GO-ABE at NSS 2015, discussing its importance in accessing patient data during emergencies. However, since Li et al.’s scheme uses bilinear mappings, it is not secure against quantum attacks. To ensure security against quantum attacks, we construct the GO-ABE scheme using the post-quantum cryptographic primitive lattices, and employ Shamir’s secret sharing scheme to meet the GO-ABE requirements.Keywordsattribute-based encryptiongroup-oriented systemsprivacylattice-based cryptography

Differential privacy is a confidentiality metric and quantitatively guarantees the confidentiality of individuals. A noise criterion, called sensitivity, must be calculated when constructing a probabilistic disturbance mechanism that satisfies differential privacy. Depending on the statistical process, the sensitivity may be very large or even impossible to compute. As a result, the usefulness of the constructed mechanism may be significantly low; it might even be impossible to directly construct it. In this paper, we first discuss situations in which sensitivity is difficult to calculate, and then propose a differential privacy with additional dummy data as a countermeasure. When the sensitivity in the conventional differential privacy is calculable, a mechanism that satisfies the proposed metric satisfies the conventional differential privacy at the same time, and it is possible to evaluate the relationship between the respective privacy parameters. Next, we derive sensitivity by focusing on correlation coefficients as a case study of a statistical process for which sensitivity is difficult to calculate, and propose a probabilistic disturbing mechanism that satisfies the proposed metric. Finally, we experimentally evaluate the effect of noise on the sensitivity of the proposed and direct methods. Experiments show that privacy-preserving correlation coefficients can be derived with less noise compared to using direct methods.