Some of the most serious security threats facing computer networks involve malware. To prevent this threat, administrators need to swiftly remove the infected machines from their networks. One common way to detect infected machines in a network is by monitoring communications based on blacklists. However, detection using this method has the following two problems: no blacklist is completely reliable, and blacklists do not provide sufficient evidence to allow administrators to determine the validity and accuracy of the detection results. Therefore, simply matching communications with blacklist entries is insufficient, and administrators should pursue their detection causes by investigating the communications themselves. In this paper, we propose an approach for classifying malicious DNS queries detected through blacklists by their causes. This approach is motivated by the following observation: a malware communication is divided into several transactions, each of which generates queries related to the malware; thus, surrounding queries that occur before and after a malicious query detected through blacklists help in estimating the cause of the malicious query. Our cause-based classification drastically reduces the number of malicious queries to be investigated because the investigation scope is limited to only representative queries in the classification results. In experiments, we have confirmed that our approach could group 388 malicious queries into 3 clusters, each consisting of queries with a common cause. These results indicate that administrators can briefly pursue all the causes by investigating only representative queries of each cluster, and thereby swiftly address the problem of infected machines in the network.
28 February 2017 in International Journal of Energy, Information and Communications
Although network management tasks are highly automated using big data and artificial intelligence technologies, when an unforeseen cybersecurity problem or fault scenario occurs, administrators sometimes directly analyze system data to make a heuristic decision. However, a wide variety of information is required to address complex cybersecurity risks, whereas current systems are focused on narrowing the candidates of information. In this study, we propose a multiagent-based data presentation mechanism (MADPM) that consists of agents operating data-processing tools that store and analyze network data. Agents in MADPM interact with other agents to form data-processing sequences. In this process, we design not only the composition of the sequence according to requirements, but also a mechanism to expand it to enable multifaceted analysis that supports heuristic reasoning. We tested five case studies in the prototype system implemented in an experimental network. The results indicated that the multifaceted presentation of data can support administrators more than the selected single-faceted optimal presentation. The final outcome of our proposed approach is the provision of a multifaceted and cross-system data presentation for heuristic inference in network management tasks.
Artificial intelligence is typically formulated as an information-processing system composed of artificial neurons, where computation is understood as recursive operations connecting inputs and outputs. However, real neural systems are materially embodied and continuously reconfigured by metabolic and physical processes, suggesting that computation cannot be reduced to fixed causal structures. In this paper, we propose a theoretical framework that captures the interplay between informational and material processes as the interaction between two computational schemes: a vertical scheme, representing fixed cause–effect relations, and a horizontal scheme, representing transformations between such relations. We show that the vertical scheme corresponds to Bayesian inference, which updates probability distributions over a fixed hypothesis space, and is consistent with the free-energy minimization principle. In contrast, the horizontal scheme is formalized as inverse Bayesian inference, which modifies the hypothesis space itself by updating likelihood structures based on experienced data. We further demonstrate that the interplay between these schemes can be expressed algebraically as a process of continuously gluing Boolean algebras. This construction yields a non-distributive orthomodular lattice, i.e., quantum logic, without invoking Hilbert space formalism. In this view, quantum logic emerges not as a static logical system but as a structural consequence of dynamically reconfiguring causal contexts. This framework provides a unified perspective in which inference is understood not only as optimization within a fixed model but also as a process that generates and transforms the model itself. It offers a formal basis for describing open-ended computation and suggests a connection to approaches such as unconventional computing and Natural Born Intelligence, where computational structures evolve through interaction with material processes. Unlike existing approaches, this framework derives quantum-logic-like structure from the continual reconfiguration of causal contexts rather than from Hilbert-space assumptions or optimization within a fixed hypothesis space.