Digital Library[ Search Result ]
Continual Learning using Memory-Efficient Parameter Generation
Hyung-Wook Lim, Han-Eol Kang, Dong-Wan Choi
http://doi.org/10.5626/JOK.2024.51.8.747
Continual Learning with Parameter Generation shows remarkable stability in retaining knowledge from previous tasks. However, it suffers from a gradual decline in parameter generation performance due to its lack of adaptability to new tasks. Furthermore, the difficulty in predetermining the optimal size of the parameter generation model (meta-model) can lead to memory efficiency issues. To address these limitations, this paper proposed two novel techniques. Firstly, the Chunk Save & Replay (CSR) technique selectively stored and replayed vulnerable parts of the generative neural network, maintaining diversity in the parameter generation model while efficiently utilizing memory. Secondly, the Automatically Growing GAN (AG-GAN) technique automatically expanded the memory of the parameter generation model based on learning tasks, enabling effective memory utilization in resource-constrained environments. Experimental results demonstrated that these proposed techniques significantly reduced memory usage while minimizing performance degradation. Moreover, their ability to recover from deteriorated network performance was observed. This research presents new approaches to overcoming limitations of parameter generation-based continual learning, facilitating the implementation of more effective and efficient continual learning systems.
Analysis of Case Scenario to Develop a System of Systems Meta-model for Ontology Representation
Young-Min Baek, Sumin Park, Yong-Jun Shin, Doo-Hwan Bae
http://doi.org/10.5626/JOK.2018.45.10.1056
Ontology is a formal and explicit specification technique that defines concepts and relationships of a system. It is utilized to establish a common knowledge base and to reduce mismatches or inconsistencies during communication. Since a System-of-Systems (SoS) is a large-scale complex system that achieves higher-level common goals by the collaboration of constituent systems, ontologies need to be established for overall SoS development and operations. In other words, refined development and communications among various stakeholders of an SoS can be achieved based on the conceptualization power of an ontology. However, in order to build an ontology effectively, SoS engineers require a systematic means to provide a guideline for domain analysis and ontology establishment. To fulfill these requirements, this study proposes a meta-model, called the Meta-model for System-of- Systems (M2SoS), which enables systematic specifications of ontologies for SoS development. M2SoS is developed based on existing studies on meta-modeling approaches in the multi-agent system domain, but M2SoS is improved to meet SoS-specific requirements by SoS case analysis.
Meta-Modeling to Detect Attack Behavior for Security
Jinho On, Yeongbok Choe, Moonkun Lee
This paper presents a new method to detect attack patterns in security-critical systems, based on a new notion of Behavior Ontology. Generally security-critical systems are large and complex, and they are subject to be attacked in every possible way. Therefore it is very complicated to detect various attacks through a semantic structure designed to detect such attacks. This paper handles the complication with Behavior Ontology, where patterns of attacks in the systems are defined as a sequences of actions on the class ontology of the systems. We define the patterns of attacks as sequences of actions, and the attack patterns can then be abstracted in a hierarchical order, forming a lattice, based on the inclusion relations. Once the behavior ontology for the attack patterns is defined, the attacks in the target systems can be detected both semantically and hierarchically in the ontology structure. When compared to other attack models, the behavior ontology analysis proposed in this paper is found to be very effective and efficient in terms of time and space.
Search
Journal of KIISE
- ISSN : 2383-630X(Print)
- ISSN : 2383-6296(Electronic)
- KCI Accredited Journal
Editorial Office
- Tel. +82-2-588-9240
- Fax. +82-2-521-1352
- E-mail. chwoo@kiise.or.kr