Abstract Artificial Intelligence, particularly machine learning (ML) algorithms, plays a crucial role in detecting cyberattacks, including anomalies and intrusions. However, machine learning models trained on imbalanced cybersecurity datasets often struggle to accurately detect minority data instances and potential threats, thereby weakening overall system security. Despite extensive research, a persistent challenge is the inadequate explanation for model predictions concerning minority data classes. This study aims to address these limitations by developing a generative AI-based approach to manage minority classes in anomaly detection, incorporating concept drift handling and explainability analysis. We introduce an over-sampling technique, CGGReaT, designed to enhance the presence of minority classes in the anomaly detection domain. Leveraging Large Language Models (LLMs) as a hybrid approach, we use pre-trained transformer-based LLM DistilGPT-2 for generating synthetic tabular data. Extensive experiments on two publicly available benchmark datasets, UNSW NB15 and CIC-IDS2017, underscore the efficacy of our proposed approach. We employed concept drift detection and adaptation techniques to maintain reliable and sustainable ML performance. To enhance interpretability, eXplainable Artificial Intelligence (XAI) methods, including SHAP and LIME, are employed to quantify feature contributions to model outputs. Extensive experiments reveal that testing ML algorithms on datasets balanced with synthetic samples generated by cGGReaT boosts the prediction accuracy on the UNSW NB15 and CIC-IDS2017 datasets, compared to classifiers tested on imbalanced datasets.

The advancement of dynamic DNA nanotechnology in the past two decades has paved the way for the bottom-up engineering of molecular devices and computers. This progress has catalyzed the emergence of molecular cybernetics, a field of study that explores versatile methodologies for integrating multiple molecular devices and computers to build adaptive, robust and plastic molecular systems. In particular, efforts are underway to engineer chemical artificial intelligence (chemical AI) capable of learning and adjusting its behaviors through interaction with the environment. This special issue assembles six research papers presenting cutting-edge reuslts on molecular cybernetics. Azuma et al. [1] have proposed a learning principle for the network of logic gates, each capable of conditioned reflex. The proposed principle opens new avenues for developing molecular machines with learning capabilities through the process of classical conditioning. Yu et al. [2] have presented a new type of surface chemical reaction network (sCRN), an emergent paradigm of molecular programming, that has less complexity in terms of the numbers of cell states and transition rules. The reduction in complexity enhances the feasibility of physically implementing such molecular networks. Komiya et al. [3] have developed a modular reaction architechture that enables rapid and significant amplification of single-stranded DNA (ssDNA) molecules in response to an input signal and experimentally demonstrated its feasibility. The reaction is B Yutaka Hori

Various visual explanation methods, such as CAM and Grad-CAM, have been proposed to visualize and interpret predictions made by CNNs. Recent efforts go beyond mere visual interpretability, aiming to enhance CNN performance through the utilization of these generated visual explanations. In this work, we propose MANet (Mixed Attention Network)—a network architecture that advances the stability of visual explanations through an adaptive feature refinement mechanism via a mixed attention module. Concurrently, the generated attention maps are harnessed to bolster network performance in image recognition tasks. Experimental findings underscore the efficacy of MANet, demonstrating improved visual stability and consistency. The proposed architecture not only surpasses baseline models in image classification and object detection tasks but also establishes a novel paradigm for synergizing visual interpretability and network performance enhancement.

A research project on chemical AI, called the Molecular Cybernetics Project, was launched in Japan in 2021 with the goal of creating a molecular machine that can learn a type of conditioned reflex through the process of classical conditioning. In this project, we have developed a learning method for the network of such learning molecular machines, which is reported in this paper. First, as a model of a learning molecular machine, we formulate a logic gate that can learn conditioned reflex and introduce the network of the logic gates. Then we derive a key principle for learning, called the flipping principle, by which we present a learning algorithm for the network to realize a desired function.

The general personality traits, notably the Big-Five personality traits, have been increasingly integrated into recommendation systems. The personality-aware recommendations, which incorporate human personality into recommendation systems, have shown promising results in general recommendation areas including music, movie, and e-commerce recommendations. On the other hand, the number of research delving into the applicability of personality-aware recommendations in specialized domains such as finance and education remains limited. In addition, these domains have unique challenges in incorporating personality-aware recommendations as domain-specific psychological traits such as risk tolerance and behavioral biases play a crucial role in explaining user behavior in these domains. Addressing these challenges, this study addresses an in-depth exploration of personality-aware recommendations in the financial domain, specifically within the context of stock recommendations. First, this study investigates the benefits of deploying general personality traits in stock recommendations through the integration of personality-aware recommendations with user-based collaborative filtering approaches. Second, this study further verifies whether incorporating domain-specific psychological traits along with general personality traits enhances the performance of stock recommender systems. Thirdly, this paper introduces a personalized stock recommendation model that incorporates both general personality traits and domain-specific psychological traits as well as transaction data. The experimental results show that the proposed model outperformed baseline models in financial stock recommendations.