Biometrics are intended to add an extra layer of security to various applications by authenticating a user based on a biometric sample taken from them. At the same time advantage, from the users' perspective, is that they are no longer required to use biometric memorizing credentials. Fingerprinting has been in use since the 19th century in determining identity. Because users' biometric templates are stored within any system, their security becomes extremely sensitive, as theft or simple copying of the biometric can compromise the identity of all users enrolled in the system. This paper presents a quantum-based biometric fingerprint encryption method. The method uses an adapted Novel Enhanced Quantum (NEQR) model to store biometric fingerprint in quantum systems and proposes quantum techniques of Confusion and Feistel algorithm to encryption it. The quantum encryption circuit was implemented and tested both in the IBM Quantum cloud and locally on the QASM quantum simulator.
As the demand for wireless communications continues to grow, securing limited spectrum resources has become an issue. Under such circumstances, Dynamic Spectrum Allocation (DSA), in which existing frequency bands are jointly used and allocated to carriers, has been attracting attention. However, the problem is that a large number of base stations and interferers causes a combinatorial explosion, which requires an enormous amount of computation time. In a previous study, this problem was viewed as a combinatorial optimization problem and formulated to be solved by quantum annealing for the time slot-specific requirements of each base station. However, each base station uses the frequency band differently, and the timeslot length requirement must also be considered. In this paper, we focus on the DSA problem of maximizing the satisfaction of the time-based requirements of each base station and describe a formulation for solving the problem using quantum annealing. These time-based requirements can accommodate the demand for bursty traffic, which was not the case in previous papers. We also present implementation results using Fixstars Amplify and discuss future prospects.
With the development of machine learning technology, as a typical application of the new retail industry, unmanned retail store has developed rapidly. Using deep learning models for retail commodity recognition provides a new way to bring efficient checkout services. In this paper, we use YOLOv5 model for building an efficient retail object recognition system. First, in order to improve the computational efficience and lower the hardware cost, we propose to use an image quality detection module for pre-processing. Blurry images are discarded for improved recognition efficiency. Second, we adopt the convolutional neural network as the back-bone network for feature extraction and object recognition. Third, we implement a real-time recognition system in the lab environment for verification of our proposed system. Experiments on RPC dataset show that the proposed system can detect the location and category of retail goods in real-time. The model performance is promising for unmanned retail applications.
The rapid proliferation of the Internet of Things (IoT) requires the development of effective methods to identify devices over a network. Although traditional machine learning and deep learning approaches have been successful in this task, they require continuous retraining when new devices join the IoT network. To address this challenge, Siamese Neural Networks (SNNs) have been proposed, demonstrating strong performance without the need for model retraining when a new device is added to the network. In this work, we further assess the reliability of SNNs in reducing the necessity for model retraining when multiple devices join the network, as well as their generalization capabilities. Our findings indicate that while performance slightly declines with an increasing number of newly added devices, it remains satisfactory overall. Furthermore, we demonstrate the generalization of SNNs by training the model on one dataset and evaluating its performance when incorporating devices from a different dataset without retraining. The results highlight the robustness of SNNs and their ability to recognize devices on a different IoT network than the one they were initially trained on, and most importantly, without retraining. Finally, we show how the quality of the initial training set can affect the generalization capabilities of SNNs.
Nowadays, Unmanned Aerial Vehicles (UAVs) tend to become an attractable platform for smart city applications, especially for environmental sensing tasks. UAV downwards view image segmentation can extract meaningful information, but the challenge lies in segmenting small-size objects. This paper proposed a reasonable method where edge features were aggregated to enhance the performance of segmentation results, together with a suitable re-weighting method to alleviate the imbalance of interested object class distribution. The effectiveness of the designed network architecture was proven by experimental results that around a 2–4% increase of IoU for aimed small-size classes was achieved. Additionally, the proposed model remains in a lightweight structure.
Tracking the surgical site in surgical images is a method that can help doctors. Therefore, we proposed a tracker model that learns by self-supervised learning for tissue organization in surgical images. We constructed a model based on unsupervised deep tracking (UDT) as a base line Our proposed model incorporates forward and backward prediction in conjunction with a Siamese network, a feature that sets it apart as a robust tracking solution. Through a series of extensive experiments, we established the effectiveness of our proposed model in comparison with existing models in the field. Our model achieved an Expected Average Overlap (EAO) score of 0.405, demonstrating significant improvements over existing models: TransT (0.274), KIT (0.223), SRV (0.293), and MEDCVR (0.302), with respective score improvements of 0.131, 0.182, 0.112, and 0.103. These results confirm that our proposed model can be helpful in performing effective tracking in surgical videos.
Neck pain may be caused by cervical bone fracture, which must be promptly detected and treated, as severe cases can lead to paralysis or even death. The diagnostic precision of radiologists in the identification of cervical spine fractures depends on the clinical manifestation of the patient. Current fracture detection accuracy among radiologists stands at only 73.98% for alert blunt traumatic patients. To address this concern, this paper presents an approach based on deep learning models that can quickly analyze CT scans and diagnose cervical spine fracture. The approach includes two stages: Stage 1 utilizes UNet-EfficientNet for CT image segmentation, while Stage 2 incorporates CrackNet-LSTM to achieve spinal injury detection. Notably, the models excel in accurately identifying fractures. Implementing these strategies with the aforementioned models yields impressive results: 99.91% accuracy for Stage 1, 94.9% accuracy for Stage 2, and a combined accuracy of 94.9% for the overall examination process. This approach significantly improves the accuracy and the efficiency, thus proving to be highly qualified in assisting radiologists and alleviating their workload in detecting cervical spine fractures.
This paper delves into the critical issue of privacy protection in video surveillance systems, emphasizing the need for comprehensive and interdisciplinary approaches. The study presents recent advancements in de-and re-identification technologies aimed at ensuring personal information protection. The mechanisms for creating high-quality anonymized images that do not compromise the original data distribution and methodologies for effective facial image restoration are explored. Special attention is given to the operational dynamics of systems under specific conditions and their management of distinct facial features. The necessity for an integrated approach in constructing AI systems in edge computing is highlighted, focusing on anonymization techniques, re-identification prevention, and data distribution maintenance. Through these research efforts, we anticipate balancing privacy protection and data identifiability, offering a solid foundation for future exploration in this field.
This study proposes a comprehensive approach combining segmentation and regression techniques for predicting anemia severity using conjunctiva images. Anemia is a prevalent health condition with a global impact. The segmentation phase employs deep learning models to accurately segment the conjunctival area, allowing precise localization of anemia-related features. The regression phase utilizes pre-trained deep-learning models to predict the severity of anemia based on the segmented images. Evaluation metrics, including Intersection over Union (IoU) for segmentation and Mean Absolute Error (MAE) and Mean Absolute Percentage Error (MAPE) for regression, assess the accuracy of the predictions. Experimental results demonstrate the potential of this approach for non-invasive anemia severity prediction, contributing to improved healthcare outcomes.
Handover authentication is a significant process which ensures access control and secure communication in multi-access edge computing (MEC). Although it has been studied for many years in the wireless communication fields, a comprehensive survey for handover authentication is still lacking in the literature. Moreover, research status and new challenges of handover authentication in the context of MEC need be investigated. This paper thereby reviews existing handover authentication schemes and presents a clear definition and classification. It also evaluates security and performance of representative handover authentication schemes. Based on the evaluation, two interesting findings are summarized and two significant research directions of handover authentication for MEC are proposed.
This paper brings some light to the importance of psychology as a concept, its initial role in human-computer interaction, and exactly how necessary it is to maintain this aspect when spearheading the future of the field. The method used is to decipher what psychology gives to Human-Computer Interaction (HCI), including the primary components, the task, user, and computer. The results of this study discuss the framework of previous research, and how it stands today considering modern sub-theories. In the end, the verdict stands that psychology as one of the feature of HCI applies primarily to authors in the United States, and very much less when it comes to European researchers, who would generally take different approaches to their research. It is necessary to have a consistent conversation on the gravity of connecting the researchers of HCI and advancements with the full extent and realm of studies that exist. One of the biggest things that this paper brings to the field is some focus on the necessity that there is for in-depth analysis of psychology as a method of approaching HCI.
Classrooms as a space for imagination and hope for students who have limitations in their education. Therefore, it is significant for students that must be taken into account and acquire urgent assistance from the teacher. Various approaches of educational technology to sustaining education are prevalent due to the impact of technology, pandemics, and educational limitation. As a consequence, the condition of teaching and learning must be changed from the original. In the past, we focused only on teaching in the classroom, but with such limitations, online learning has an important role. One of these approaches, Blended Learning, is known as combining online and face-to-face learning with the support of educational technology. It is becoming more prevalent. Therefore, the teacher's perspective must be able to manipulate blended learning by linking online and face-to-face learning environments with educational technology support. To improve students' achievement for educational opportunity expansion schools for a better quality of education.
If large-scale quantum computers capable of running quantum algorithms are developed, it is expected that powerful quantum cryptanalysis will become possible. To establish secure quantum-resistant encryption systems, it is necessary to evaluate the quantum security of cryptographic algorithms. As a result, various quantum cryptanalysis research studies have been published, and implementation techniques that minimize the resources required for analyzing encryption have been proposed. In this paper, we present an efficient quantum circuit implementation of Classic McEliece, which is one of the candidate algorithms in the NIST Post-Quantum Cryptography Standardization Round 4. We optimize the encoding, and decoding operations of Classic McEliece on quantum circuits, with a specific focus on binary field arithmetic, linear operations, and the Berlekamp-Massey decoding quantum circuit.
E-mail is mainly used by organizations such as businesses and schools for efficient work processing. Because it handles various tasks, most of the information related to the task is stored in e-mail. If important documents of a company are leaked due to mis-transmission of the e-mail, there is a possibility that the leaked documents may be abused. In preparation for such a case, the sender may cancel the transmission of the e-mail using the ‘cancel transmission’ function. However, if the recipient can restore the canceled e-mail, the sender cannot completely block the mis-transmission of the e-mail. In this paper, we analyze the e-mail storage structure stored in the e-mail data file and the storage structure of the unsent e-mail remaining in the file and attempt to restore the email.
Generative AI technology is being applied in various fields. However, the advancement of these technologies also raises cybersecurity issues. In fact, there are cases of cyber attack using Generative AI, and the number is increasing. Therefore, this paper analyzes the potential cybersecurity issues associated with Generative AI. First, we looked at the fields where Generative AI is used. Representatively, Generative AI is being used in text, image, video, audio, and code. Based on these five fields, cybersecurity issues that may occur in each field were analyzed. Finally, we discuss the obligations necessary for the future development and use of Generative AI.
Traditional education methods, having said, lack motivation and self-regulation. The 5G and 6G technology have improved education; one such is gamification. Gamification applies gaming mechanisms to non-gaming platforms. Sadly, the use of gamification is still limited, especially in Malaysia. Gamification affects user experience, so it is intriguing to see if it can solve education problems. A study was conducted to identify elements of gamification. A systematic literature review (SLR) determines the gamification elements. Results show the list of identified 30 elements as guidelines for developing gamified teaching in Computer Science Courses. Thus, it is hoped that the improved teaching and learning process can affect students' performance.
Multiplicative complexity is one of the important properties for the efficiency of S-box. In this paper, we present a new method to find S-box circuits with optimal multiplicative complexity. Our method uses the early abortion technique to speed up finding the canonical forms of the S-box circuits. We adopted a method of grouping AND gates and searching the grouped parts. One or more component functions can be generated by searching each part, and an S-box can be generated by searching all parts. This allows a wider range of S-boxes to be investigated. We believe that our method will help S-box designers to generate efficient S-box circuits.
There are various techniques (String Search, Signature, List Traversal, Kernel Object, etc.) to perform memory forensics. Among them, Kernel Object-based memory forensics techniques that utilize the object structure of the kernel are considered the most reliable. Kernel Object-based memory forensics techniques require prior knowledge of the object structure of the operating system kernel used in the memory dump. However, reverse engineering the kernel for a vast number of operating system versions and architectures to identify the object structure is labor- and time-consuming. To solve this problem, academic researchers have developed methods to efficiently identify the structure of various kernel objects. Various studies have been conducted to identify key features that kernel objects leave in memory, or to use automation technology. We will review these works and discuss what further research can be done and the challenges that need to be considered.
Conversational interfaces allow users to experience artificial intelligence (AI) services through text or voice conversations. One common form of a conversational interface is a chatbot, which can be scenario-based or large language model (LLM)-based. A scenario-based chatbot generates a response within a predefined scenario for a user query on a specific domain or topic. The chatbot's response for recommendation is processed in conjunction with a separate algorithm. A LLM-based chatbot generates a response through a pretrained model to a user query on a wide range of topics. In this process, the LLM-based chatbot's response takes the form of a kind of recommendation, which is different from the existing recommendation services. To look at the issue more comprehensively, this paper examines recommendation-style system responses of a LLM-based chatbot with the principles of AI ethics. Several examples are shown where the chatbot's responses are modified according to principles of AI ethics.
This paper introduces a reinforcement learning-based framework designed to tackle dynamic pricing challenges in e-commerce. Prior research has predominantly concentrated on algorithm selection to enhance performance in dense data scenarios. However, many of these models fail to robustly address sparse data structures, such as low-traffic products, leading to the ‘cold-start’ problem [4]. Through numerical analysis, our framework offers innovative insights derived from the design of the reward function and integrates product clustering with pre-trained learning to mitigate this issue. As a result of this optimization, the performance of predictive models on sparse data is expected to see substantial improvement.
