With the increasing demand for Bluetooth devices, various Bluetooth devices support a power-saving mode to reduce power consumption. One of the features of the power-saving mode is that the Bluetooth sessions among devices are temporarily disconnected or are close to being disconnected. Prior works have analyzed that the power-saving mode is vulnerable to denial of sleep (DoSL) attacks that interfere with the transition to the power-saving mode of Bluetooth devices, thereby increasing its power consumption. However, to the best of our knowledge, no prior work has analyzed vulnerabilities or attacks on the state after transitioning to the power-saving mode. To address this issue, we present an attack that abuses two novel vulnerabilities in sleep mode, which is one of the Bluetooth power-saving modes, to break Bluetooth sessions. We name the attack Breaktooth. The attack is the first to abuse the vulnerabilities as an entry point to hijack Bluetooth sessions between victims. The attack also allows overwriting the link key between the victims using the hijacked session, enabling an arbitrary command injection on the victims. Furthermore, while many prior attacks assume that attackers can forcibly disconnect the Bluetooth session using methods such as jamming to launch their attacks, our attack does not require such assumptions, making it more realistic. In this paper, we present the root causes of the Breaktooth attack and their impact. We also provide the technical details of how attackers can secretly detect the sleep mode of their victims. The attackers can easily recognize the state of the victim's Bluetooth session remotely using a standard Linux command. Additionally, we develop a low-cost toolkit to perform our attack and confirm the effectiveness of our attack. Then, we evaluate the attack on 17 types of commodity Bluetooth keyboards, mice and audio devices that support the sleep mode and show that the attack poses a serious threat to Bluetooth devices supporting the sleep mode. To prevent our attack, we present defenses and their proof-of-concept. We responsibly disclosed our findings to the Bluetooth SIG. We also released the toolkit as open-source.

The reuse of third-party code, such as open-source software (OSS), enhances software development efficiency but may introduce vulnerabilities that pose significant risks to systems. This paper focuses on known vulnerabilities originating from reused code, referred to as “code clone” (CC), with the specific term vulnerable CC used to denote vulnerable fragment. Previous studies only detect vulnerable CCs that are almost exactly matched or within a limited scope in the inspected software. In this paper, we developed SHERRY, a precise approach to detecting vulnerable CCs. It enables the detection of vulnerable CCs that are not precisely matched by converting the function code into a fine-grained set of features consisting of line-by-line elements. For scalability, SHERRY reduces comparisons and calculations similarity using logical operations. Furthermore, We analyzed 50 high-profile OSS projects, tracking vulnerable CCs detected by SHERRY and examining how developers manage them. SHERRY improved recall by over 10% and accelerated processing time 17-fold without limiting scope in a comparison experiment with existing techniques using the same 10 OSS. Our measurements also revealed 87 vulnerable CCs in 22 OSS projects, and more than half of them were comparable to the most dangerous software weakness type. We finds that there are three causes of why vulnerable CCs remain in OSS repositories. Ultimately, we conclude with practical suggestions to prevent the propagation of vulnerable CCs in the OSS ecosystem.

Large language models (LLMs) have demonstrated exceptional capabilities in generating text, images, and program code, driven by extensive training on large-scale datasets. Among LLMs, ChatGPT has emerged as a prominent example, with over one hundred million users worldwide. Previous studies have highlighted its ability to generate program source code for symmetric-key ciphers such as AES, CHAM, and ASCON. This study investigates the implementation of cryptanalytic program source code for lightweight authenticated encryption scheme ASCON and lightweight block cipher SLIM using ChatGPT, aiming to evaluate its potential and limitations in the field of cryptography.

We examine the security of AES-based authenticated encryption schemes, including the AEGIS family, Tiaoxin-346, Rocca and Rocca-S. Existing studies evaluated the security against forgery attacks, focusing on state collisions in the encryption phase. These studies estimated the lower bounds for the number of active S-boxes by a byte-wise search. However, this approach might underestimate these bounds since it might include invalid characteristics. In this paper, we conduct a bit-wise evaluation of the AEGIS family, Tiaoxin-346, Rocca, and Rocca-S against forgery attacks based on state collision by Boolean satisfiability problem (SAT) tools. This approach enables us to derive tighter bounds for the minimum number of active S-boxes. In addition, for AEGIS-128L, Tiaoxin-346, and Rocca, we incorporate values of differential distribution tables of S-boxes to obtain the exact differential characteristics probability as a direct indicator of forgery attacks on AEGIS-128L, Tiaoxin-346, and Rocca. Additionally, we conduct a bit-wise evaluation of differential distinguishing attacks on the finalization phase of the AEGIS family, Tiaoxin-346, Rocca, and Rocca-S. These results reveal that AEGIS-128L cannot claim 256-bit security against forgery attacks even with a 256-bit tag. Furthermore, for the first time, we perform a security evaluation against forgery attacks exploiting tag collisions in the tag generation phase. Finally, we reveal that the finalization phase cannot be distinguished from the pseudo-random permutation under differential attacks.

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.

Multi-signatures are protocols used when multiple signers wish to produce a joint signature on the same message. They are principally used in blockchains. In this paper, we propose a multi-signature by extending the signature Commutative Supersingular Isogeny based Fiat-Shamir (CSI-FiSh) with Sharing-friendly Keys (CSI-SharK) signature scheme to a multi-signature. We use a non-interactive zero-knowledge proof system to make sure the signers make the required computations. Both the Key Aggregation and MultiSigning use a sub-protocol to allow signers to aggregate elliptic curves in a round-robin manner. We prove the resulting multi-signature protocol is actively secure in the random oracle model (ROM) by using the Double Forking Lemma Technique. We compare the resulting scheme to the most efficient isogeny-based threshold protocol and show that while our protocol lacks the robustness property, the multi-signature is more advantageous in a situation where a group of signers would want to recurrently output signatures. We also compare our multi-signature to two state of the art lattice-based multi-signatures, and show that by carefully choosing the parameters, our multi-signature is smaller by a significant factor. Finally, we implement our multi-signature scheme using C.

When Protective DNS (PDNS) detects a malicious attempt to resolve a domain name, it blocks the connection to the malicious domain by rewriting the DNS record. PDNS protects a user from a malicious domain; thus, it has been promoted in recent years. However, it is unclear to users which domains can be blocked, and unclear to providers how to implement and operate PDNS. To address the ambiguity of PDNS, we conducted a survey regarding its blocking coverage, lifecycle, blocking of popular domains, and usability. Our investigation revealed that the blocking rate varies widely among PDNSs, with some having a maximum of approximately 55.32% while others have a blocking rate of less than 1%. We also explored the lifecycle of each PDNS through continuous observations over a 32-day period and found that approximately 80% of all PDNSs blocked malicious domains on the same day that the domain was listed on the blocklist. Furthermore, we found that up to 84.54% of the blocked malicious domains were unblocked over time, and the number of days until unblocking ranged from 3.20 to 9.34 days. In addition, we evaluated the false block rate for each PDNS for popular domains and the usability of each PDNS's services. Finally, we provide recommendations for both PDNS users and providers based on the results of the investigation and analysis.

In dynamic malware analysis, where malware is executed and its behavior analyzed, screenshots (referred to as “analysis screens”) can be used to capture the screens displayed within the analysis environment. These analysis screens can provide information on malware behavior, such as launched applications and end-user deception techniques, thereby contributing to the efficiency of dynamic malware analysis systems. They also have potential applications in expanding analysis systems and improving end-user security awareness through education. However, there has been no comprehensive investigation into the information obtainable from analysis screens, and their potential applications have not been clearly demonstrated. In this study, we conducted an investigation through coding of 3, 590 analysis screens included in 211 analysis reports covering a total of 93 malware families, with the aim of organizing the information obtainable from analysis screens. As a result of this investigation, we identified malware-related information and end-user deception techniques obtainable from analysis screens. Additionally, by comparing analysis screens with logs, we demonstrated the existence of information that can be more easily obtained from analysis screens and examined their potential applications. We believe our findings contribute to the development of dynamic analysis systems and educational guidelines for analysts and end-users in the future.

This paper proposes an account recovery framework with location-based dynamic questions (ARF-L) to enhance account recovery processes. It leverages users' location histories to estimate visited places, frames questions by selecting suitable points from these locations, and then generates questions with corresponding choices. Preliminary experiments included a survey with 1,000 participants to gauge the social acceptability and prerequisites for generating such questions. A smaller scale manual question generation experiment with eight participants demonstrated high accuracy in responses and a favorable attitude towards our framework. Building on these insights, we proposed question generation algorithms and validated them on 38 participants, achieving responses with an 86% average accuracy rate. This study underscores the potential for using location history in security practices, although it also highlights challenges such as refining question generation algorithms. Our future efforts will address these challenges and ensure comprehensive reliability and applicability across different contexts of our approach.

Various models for evaluating anonymity have been proposed so far. Among them, k-anonymity is widely known as a typical anonymity measure, guaranteeing that at least k individuals in a database have the same values. Unfortunately, it is difficult to create highly useful anonymized data satisfying k-anonymity for high-dimensional data because of the curse of dimensionality. Several approaches relaxing k-anonymity have been proposed, such as k^m-anonymity, to overcome the problem. On the other hand, we have another privacy protection metric, developed by Dwork et al., called differential privacy. However, the complete protection index for differential privacy, i.e., the level of noise that can satisfy the desired privacy, has not been clarified. This paper shows relationships between k^m-anonymity and differential privacy under sampling, proposed by Li et al., that is, a weak notion of differential privacy. Numerical experiments are then performed to give relations among the parameters of k^m-anonymity and differential privacy under sampling. These experiments also show relationships between k-anonymity and k^m-anonymity as k-anonymity is a special case of k^m-anonymity in some sense.

We developed a support system for the exhibits of Edo-era sugoroku games in museums. This system assists visitors to learn and play old sugoroku games. Visitors can play the games according to instructions displayed on a monitor and a projector without prior knowledge of sugoroku games. The system provides visitors with a die that is equipped internally with a microcomputer. Since the system detects the value of a die automatically, visitors can easily progress through the game. In addition, the system records the progress of games and the operations performed by visitors, which can be used to analyze visitors' behaviors. The system was used at a special exhibition of the National Museum of Japanese History for eight weeks. We conducted a three-day evaluation with visitors and confirmed that computer-aided Sugoroku raised the interest in Edo-era Sugoroku of over 90% of the visitors. Moreover, the analysis of the log data confirms that the functions of the system can be effective in museum exhibits.

Power and energy consumption are becoming the most important restrictions on driving computer systems. Especially, HPC systems still require more performance despite their limited power supply, and it is necessary to explore methodologies to enhance their performance within this limitation. Approximate computing (AC) is a promising technique for optimizing the trade-offs among application execution performance, application output correctness, and power consumption during execution. However, applying AC sometimes causes unexpected effects on the output and performance due to cache behavior, compiler optimization, and so on. This paper discusses dynamic approximate computing methods, which enable changing data precision in applications at runtime, for more flexible and effective approximate computing (AC). This paper also evaluates the benefits and drawbacks of dynamic AC with some high-performance computing (HPC) applications and workloads.

One of the success factors of business process outsourcing (BPO) is a comprehensive and in-depth understanding of the business processes that are outsourced. However, such business processes are often undocumented, and discovering them is difficult and time-consuming for BPO service providers. Previously, we introduced an approach for business process discovery that uses swim lanes to recognize that different parts of the process may be performed by different parts of an organization. It generates the business process from an event log and user information extracted from an existing system. This paper examines whether the approach aids the engineers and consultants of the providers who need to investigate business processes of organizations (i.e., their customers of which they do not have sufficient knowledge) discovering the undocumented business processes of the target organization. To do so, we surveyed an industrial workflow system from which we collected data about 2,000 events and 269 users for a two-year period of the system's operation. We conducted a study on the case through document analysis and expert interviews. Our evaluation showed that 46% of the process patterns were undocumented, and 36% of these were invalid patterns. The approach is a valuable process visualization for identifying knowledge that is not documented nor recognized even by experts in organizations.

State machine replication (SMR) is a replication technique that ensures fault tolerance by duplicating a service. Geo-replicated SMR is an enhanced version of SMR that distributes replicas in separate geographical locations, making the service more robust against large-scale disasters. Several geo-replicated SMR protocols have been proposed in the literature, each tailored to specific requirements; for example, protocols designed to reduce latency by either sacrificing a part of their fault tolerance or limiting the content of responses to clients. However, this diversity complicates the decision-making process for selecting the best protocol for a particular service. In this study, we introduce a latency estimation model for these SMR protocols based on the communication patterns of the protocols and perform simulations for various cases. Based on the simulation results and an experimental evaluation, we present five selection guidelines for geo-replicated SMR protocols based on their log management policy, distances between replicas, number of replicas, frequency of slow paths, and client distribution. These selection guidelines enable determining the best geo-replicated SMR protocol for each situation.

The recruitment of perpetrators for special frauds and robberies under the pretense of “high-paying part-time jobs” and the methods used to deceive victims into relinquishing money or valuables have emerged as critical issues on social media. Current approaches have been proposed to detect illegal and harmful content such as analyzing contextual features like slang in posts and identifying suspicious accounts through network graph analysis. However, their efficacy in addressing this threat remains unverified. This study seeks to elucidate the characteristics of these threats and evaluate the effectiveness and limitations of conventional detection methods. To achieve this, we collected and analyzed posts from X (formerly Twitter) that appeared to be associated with attempts to incite online fraud or criminal activities. The analysis revealed that posts likely associated with criminal activities often included specific place names and job details, suggesting that conventional detection methods may be partially effective in identifying such content. Conversely, posts potentially related to online fraud frequently employed language referencing the personal traits of individuals targeted for recruitment into high-paying part-time jobs. Additionally, these posts commonly utilized mentions and emojis, distinguishing them from those associated with criminal activities. This study's primary contribution is in demonstrating the scope and challenges of the previously studied methods and provides knowledge that can be applied to the automatic detection of posts on social media recruiting people for high remuneration.

Software supply chain security has relied upon layered protective measures, such as fuzzing, code signing, and secure coding, to protect against unintentional vulnerabilities and intentional tampering. Regrettably, attacks, such as Solar Winds and Log4Shell zero-day, demonstrated that current protections are insufficient. As a result, several projects have emerged, aimed at providing rigorous protections, focusing largely on dependency management, code signing, and binary file tracking. A common approach adds developer identity within the code signing ecosystem, establishing a chain of trust between developers and code-signing keys. However, these solutions depend upon external identity providers performing authentication correctly, leaving potential for account hijacking and other identity-based attacks. Mitigation is offered via monitoring and auditing, but relies on other parties to actively monitor for anomalies. In this paper, we propose and evaluate a FIDO-based extension to the Sigstore system, which would embed authentication data into the signing process, providing end-users with added identity assurance, complementing Sigstore's key-to-identity mapping. By providing attestation information to increase authentication strength, we can potentially issue longer lifetime developer certificates, reducing the overall number, for a more scalable system. We also perform a basic evaluation to demonstrate that our improvements can be implemented feasibly with minimal changes to Sigstore.

This study addresses the challenge of identity management for researchers who change their institutional affiliations, aiming to enable efficient and continuous use of research data across educational and research organizations. Although many organizations are developing their own research data management systems, researchers often lose access to previously available datasets when they move to new institutions. To tackle this issue, we propose a cross-organizational identity proofing system that maintains the assurance levels of IAL2/AAL2 (as defined by NIST SP 800-63) and completes the ID transfer entirely online. The core idea is to link pre- and post-move IDs by verifying the user's identity with researcher ID numbers, ORCID, and government-issued credentials. Specifically, we employ the Japanese Individual Number Card (commonly known as the “My Number Card”) to achieve robust identity verification, allowing seamless ID linkage among Identity Providers with high assurance levels. The system is implemented in the Orthros development environment—an ID linkage service provided by the National Institute of Informatics (NII)—using xID Co.'s API for Individual Card integration. Our approach is expected to mitigate ID management issues arising from researcher transfers and to enhance the efficiency of research data utilization across organizations.

Online disinformation often provokes strong anger, driving social media users to spread it; however, few measures specifically target sharing behaviors driven by this emotion to curb the spread of disinformation. This study aimed to evaluate whether digital nudges that encourage deliberation by drawing attention to emotional information can reduce sharing driven by strong anger associated with online disinformation. We focused on emotion regulation to encourage deliberation. Digital nudges were designed to display an emotion pie chart about disinformation and emotion regulation messages. In particular, distraction and perspective-taking nudges were found to encourage deliberation in anger-driven sharing. We assessed their effectiveness with existing nudges mimicking platform functions. The results showed that all digital nudges significantly reduced the sharing of disinformation, with distraction nudges being the most effective. These findings suggest that digital nudges addressing emotional responses can serve as an effective intervention against the spread disinformation driven by strong anger.

Due to the recent popularity of sports cycling, many people are interested in cycling tourism. Notifying novices of irregular cycling forms is helpful to avoid muscle fatigue. However, the conventional technologies for detecting irregular cycling forms by inertial measurement unit (IMU) alone face significant limitations in accurately capturing cycling movements. Moreover, simple threshold-based methods fail to account for individual differences among cyclists and fatigue-induced variations. Hence, we propose a novel AI architecture for detecting irregular cycling forms. The proposed architecture combines data from IMUs and 2D Light Detection and Ranging for more accurate detection of irregular cycling forms. Our proposed architecture achieved 0.900 in accuracy, 0.892 in precision, and 0.841 in recall for an F1-score of 0.862, demonstrating the effectiveness of our approaches. Compared to models using only IMUs data, our proposed architecture achieved a significant improvement in precision, increasing by 10.0 points.

The Seat Arrangement Problem is the problem of finding a desirable seat arrangement for given preferences of agents and a seat graph that represents a configuration of seats. In this paper, we consider the decision problems of determining if an envy-free arrangement exists and an exchange-stable arrangement exists, when a seat graph is an l × m grid graph. When l=1, the seat graph is a path of length m and both problems have been known to be NP-complete. In this paper, we extend it and show that both problems are NP-complete for any integer l ≥ 2.

Online communication such as Slack is useful, but has the problem that it is difficult to understand the workplace situation and the status of employees, which are easy to understand in a traditional face-to-face environment. To solve this problem, we propose a method to visualize communication and identify lonely users. For visualizing online communication, we define two indices: Contribution Level, representing a user's contribution to communication, and Adjacency Level, representing the strength of relationships between users. For identifying and clearly visualizing lonely users, we apply clustering to the generated graph. To verify the validity of the proposed method, we analyzed data from our laboratory's Slack and conducted a survey using the UCLA Loneliness Scale and a subjective evaluation questionnaire for laboratory members. As a result, we concluded that users who communicate less online do not necessarily feel lonely, but it was also revealed that users with low scores on the UCLA Loneliness Scale had a significant difference in Adjacency Level compared to other users. It was suggested that the proposed system may be able to discover potential risks of loneliness.

Modeling multiple-choice behavior is an important research topic in the field of social sciences, including consumer behavior research. A model that has been studied for a long time is the multinomial logit model; however, it has been shown to have problems, such as the need to set different parameters for each choice used as the objective variable. Furthermore, the number of parameters increases exponentially as the number of choices considered in the model increases. This results in unstable parameter estimates and makes variable selection difficult. In particular, when using discrete variables consisting of many categories as explanatory variables, it is necessary to set dummy variables for each category, and the effect of multicollinearity may further destabilize parameter estimates. To address these issues, we focused on cases where explanatory variables are composed of discrete variables and converted the categories contained in the discrete variables into numerical variables using the Weight of Evidence. Furthermore, we attempted to reduce the number of parameters in the model. In addition, by including only variables having a strong causal relationship with choice behavior based on the Information Value and correlation coefficients to the variable list, we proposed improvements to the model to simplify variable selection in the multinomial logit model. We then report the results of applying conventional methods and the method proposed in this study to real data, demonstrating the effectiveness of the proposed method.

I have been developing and improving a pictograph (emoji) chat that allows people from other countries to communicate in simple conversations using only emojis. As a result of numerous experiments, the communication abilities of the emoji chat in simple conversations with people from other countries and the functions necessary to support it have become clear and are described as follows: (1) In a chat method in which one line of emojis is sent alternately, the other person may understand 90% of the messages in a simple conversation such as icebreaking, both face to face and remotely, using 500 types of emojis. The same chat method, using emoji fixed phrases from a cooperative online role-playing game (RPG), achieved a similar level of understanding (90%). (2) For both face-to-face and remote conversations, when the subject is uniquely established, the upper limit of this method is close to 2.4 utterances per minute on average. (3) In the case of icebreaking conversations, appropriate categories should be created. These include subjects, 5W1H, time relationships (e.g., past, present, and future), weather, facial expressions (e.g., smiley faces), verbs, adjectives, a right-pointing arrow, places and vehicles, nouns (e.g., everyday objects, animals, food, and games), alphabets, numbers, marks, and time. Moreover, emojis that can be understood by people from other countries are necessary. (4) In terms of support functions, the history tab and quotation function are used approximately 30% of the time, and fixed phrases are used 74% of the time. As the number of emojis per minute in the system without these support functions is significantly lower (approximately half) than that in the system with support functions, it can be assumed that these support functions work effectively.

This paper shows that the puzzle named UDLR_Modify is PSPACE-complete by constructing a reduction from the PUSH-1F problem, which is a variation of Sokoban.

Reactive programming (RP) is a programming paradigm that abstracts time-varying values to describe data flows from input to output in a declarative manner. However, in distributed RP, update propagation takes time and introduces delays. Accordingly, there are various ways to reconcile execution results under delay. This study explores SignalJ, an RP language leveraging timestamps for consistent execution results. This method easily realizes persistent time-varying values that allow retrogression of computation results into the past. However, in situations where the signal network changes dynamically, inconsistencies arise unless update propagation through the network is treated as an atomic process. To address this, we propose an algorithm that precisely blocks update propagation across the entire connected data flow until the ongoing propagations are finished and the switch process for signal networks is completed. This is necessary to realize an atomic process in a situation wherein each node of the data flow only possesses knowledge of its neighbors. This maintains the consistency of update propagation during network changes. We evaluated the execution time of this algorithm. Even though the blocking overhead is not small in cases where there are long paths from sources to sinks in the network, this overhead can be reduced by introducing a caching mechanism into persistent time-varying values.

Service identification from TLS-encrypted IP flows is useful in several ways, such as providing zero-rating service. In the case of TLS 1.2 or lower, the SNI fields are not encrypted and can be analyzed for service identification, then, for flows with TLS 1.2 or less, a method was proposed to identify the service based on SNI occurrences. The method first investigates the relationship between the SNI occurrences and the services being accessed. It then identifies the service from the IP flows based on the SNI occurrence using Bayesian inference. In this paper, we focus on this method, and discuss its improvement in identification accuracy. We then show that the giving up identification caused by the absence of the SNI occurrence pattern in the previously created database is one of the main reasons for the decrease in accuracy. To solve this problem, we propose to exclude SNIs from information basing identification according to entropy. The proposed method excludes SNIs in the order of increasing the amount of relative entropy reduce by SNIs until the identification is not given up. We compare the accuracy of the existing and proposed methods and show that the proposed method improves the identification accuracy.

Incorrectness Separation Logic (ISL) is a proof system designed to automate verification and detect bugs in programs manipulating heap memories. In this study, we extend ISL to support variable-length array predicates and pointer arithmetic. Additionally, we prove the relative completeness of this extended ISL by constructing the weakest postconditions. Relative completeness means that all valid ISL triples are provable, assuming an oracle capable of checking entailment between formulas; this property ensures the reliability of the proof system.

General purpose computing on graphics processing units (GPGPU) has an execution model in which the number and type of parallel tasks are managed by the CPU, making it difficult to execute fine-grained parallel programs efficiently with nested parallel tasks at a nonhomogeneous granularity. This work addresses this problem by efficiently executing fine-grained parallel programs by managing parallel tasks on the GPU using a fast memory allocation mechanism. As a preliminary implementation, this work proposes a method for splitting the computation in a fine-grained parallel fork-join program at the fork point and allocating each computation to the GPU memory as a parallel task. In addition, kernel fusion, parallel task reuse, and parallel throttling are explored as optimization methods for the proposed method. This work implements a fine-grained parallel fork-join program in CUDA and investigates its scalability and execution speed to evaluate the feasibility and performance of the proposed method.

Equivalence relations for process algebras have been extensively studied due to their importance in analyzing the properties of concurrent systems. While bisimulation and bisimulation-based congruences are common, the ambient calculus (AC)-a formalism specialized for representing dynamically changing hierarchical structures-typically employs contextual equivalence (CE). In our previous work, we presented an example where CE fails to distinguish between two AC processes with distinct properties, and we proposed an alternative congruence relation capable of this distinction. Separately, we proposed the Synchronized Ambient Calculus (SAC), a language designed to reduce the occurrence of grave interference, an AC-specific type of synchronization failure. In this paper, we demonstrate that the aforementioned example can be expressed more concisely using SAC, and we propose a new congruence relation for SAC that successfully distinguishes between them.

In this research, we propose a method to efficiently retrieve instance segmentation models that perform well on a specific image. With the advancement of machine learning and computer vision technologies, image instance segmentation has found widespread applications in industrial domains such as quality inspection, automated manufacturing, medical diagnosis, and autonomous systems. However, obtaining optimal results requires selecting an appropriate model from numerous candidates, which demands substantial time and computational resources as each model must be individually tested on target images. Our approach addresses this challenge by embedding candidate models and target images into high dimensional vector spaces and leveraging these representations to predict performances, without actual inference. We combine performance prediction with the Learning to Rank technique to accurately model the relative performance relationships among candidate models. The proposed method effectively identifies appropriate models for a specific image while eliminating the computational burden, offering a practical solution for real-world applications where computational efficiency is critical. To evaluate our method, we constructed an instance segmentation dataset containing 20 diverse categories, each with 20 images featuring varying object shapes, colors, and sizes. Experimental results using cross-validation demonstrate that our approach significantly outperforms baseline methods.

The development of natural language understanding (NLU) models for dialogue systems necessitates the collection of a large volume of user utterances as training data, which requires significant human effort. To improve the efficiency of data collection, we develop a novel active utterance collection framework that leverages dialog scenes, which are the states of the dialog manager in the system, to actively control the data collection process. The key idea of the proposed method is to identify dialog scenes where the current NLU model performs worse and collect more data instances in those scenes to efficiently improve the model's performance. To estimate the performance of the NLU model on each dialog scene, we propose two strategies to generate validation data, including a method that uses large language models (LLMs). Empirical evaluations on the Schema-Guided Dialog dataset indicate that the proposed method can improve the efficiency of data collection in scenarios where a substantial labeled validation dataset is available. However, its efficacy diminishes in settings with practical constraints that limit the availability of validation data. These findings underscore the potential of the proposed approach, which opens new avenues for future research in practical methods for enhancing the efficiency of data collection in dialog systems development.

Fault tolerance in autonomous driving systems is important for ensuring safe operation even under hazardous conditions, and is therefore a key consideration for deployment. In an autonomous driving system with redundant devices, it is necessary to maintain consistency across the system regarding which devices should handle processing at any given time. This paper introduces a new leader election algorithm tailored for autonomous driving systems and proposes a redundant system architecture that leverages this algorithm. To ensure practicality in autonomous driving scenarios, the algorithm is designed to handle any single failure, including network link crashes. Given the safety-critical nature of the environment and the inherent challenges in testing distributed algorithms, formal verification methods were employed to ensure both safety and liveness. In designing the redundant system, we propose a mechanism that enables safe stopping, even under constraints of limited computational resources. Experimental results show that the leader election algorithm operates within reasonable time frames in a simulated environment, and that the proposed redundant autonomous driving system successfully achieves in-lane stopping in scenarios requiring safe stops, such as navigating curved road segments.

The global online video game industry has become a more than one hundred billion dollar market. Cheating is one of the major issues in multiplayer online video games. Cheat prevention is challenging because of the unique threat model of online games. Because the cheaters physically possess their own game computers, they can install cheat tools at the user level and OS and can access the physical DRAM bus to tamper with the game process. This paper aims to efficiently prevent memory tampering cheats by leveraging a trusted execution environment (TEE), Intel Client SGX. It provides strong integrity protection against memory tampering. Unfortunately, its strong protection comes with a severe size limitation of the available enclave memory. Running the entire game process inside the enclave exceeds the memory limit. This paper presents Gangi, a library that efficiently protects the integrity-sensitive game state that exceeds the available enclave memory. Gangi places the game state outside the enclave to reduce memory consumption inside the enclave while ensuring the integrity of the game state by hash-based validation. Our benchmarking results show that a Gangi-protected game outperforms the EPC swapping approach with the entire game state being inside the enclave.

This paper introduces the SAREF4ECHONET ontology, an extension of the Smart Applications REFerence ontology (SAREF) for the international standard IEC:62394, which defines the ECHONET Device Objects specification—a widely used standard for modeling smart home appliances in Japan. The proposed extension bridges the semantic gap between SAREF and ECHONET by providing a comprehensive ontology that aligns ECHONET-based devices, resources, and functionalities with SAREF's generic semantic framework. SAREF4ECHONET enhances the semantic interoperability of ECHONET-based appliances and paves the way for the automatic achievement of interoperability between ECHONET Lite and the ECHONET Lite Web API with other systems through semantic technologies.

Currently, active research on learning analytics is being conducted at research institutions worldwide. However, relying solely on objective data derived from the accumulation and analysis of learning data may not discern internal factors influencing learners' decision-making in their academic pursuits. Therefore, a combination of subjective data collection and analysis is essential for learning and educational support. This study sought to examine how the provision of learning-analytics-based feedback can stimulate a change in learners' consciousness and action to improve their learning in discrete mathematics class. Objective and subjective data were collected simultaneously from e-learning and motivation tests, respectively, and analyzed using machine learning. Thereafter, the results were provided as relative evaluation feedback to the learners. To confirm its effectiveness, we provided feedback in the class for three years. To investigate the effects of feedback on student motivation, we compared the results for the motivation scales of the Motivated Strategies for Learning Questionnaire (MSLQ) survey conducted before and after the feedback. The results confirmed that for the factors of MSLQ survey that had significant effects on learning outcomes, the differences in the mean subscale scores before and after feedback were positive for “value of learning, ” “self-efficacy, ” “external goal-orientation, ” “possession of skills, ” and “internal goal-orientation” and only negative for “test anxiety”. Teachers can utilize this feedback to stimulate a change in their consciousness and action to improve their learning, and, subsequently, their learning outcomes in discrete mathematics.

Chatbots, as conversational agents designed to deliver information on demand, have become increasingly prevalent across various domains, including consumer services, business operations, and local government initiatives. This study documents the development of a LINE chatbot system integrated with a laboratory website using Amazon Web Services (AWS) to efficiently disseminate research findings. The focus is on addressing practical challenges encountered during development, such as system integration and API management, while deriving strategies to enhance user interaction and engagement. Comprehensive user evaluations were conducted to assess our system and collect actionable feedback for iterative refinement. This paper highlights the practical insights and lessons learned from these real-world applications, offering generalized knowledge that supports broader implementation and adaptation in similar technological and organizational contexts.