The design, evolution and reuse of software system architectures are always important research areas in softwareengineering. In this paper, we first propose the concept of a multi-level orthogonal software system arc...
详细信息
The design, evolution and reuse of software system architectures are always important research areas in softwareengineering. In this paper, we first propose the concept of a multi-level orthogonal software system architecture, then describe a methodology of evolving and reusing this software architecture. Furthermore, we apply these concepts and methods to practical work, and are confident that they work well in reducing the complexity of software evolution (especially for large-scale software) and in enhancing the reuse rate.
Edge intelligence (EI) integrates edge computing and artificial intelligence empowering service providers to deploy deep neural networks (DNNs) on edge servers in proximity to users to provision intelligent applicatio...
详细信息
Edge intelligence (EI) integrates edge computing and artificial intelligence empowering service providers to deploy deep neural networks (DNNs) on edge servers in proximity to users to provision intelligent applications (e.g., autonomous driving) for ubiquitous Internet of Things (IoT) in smart cities, which facilitates the quality of experience (QoE) of users and improves the processing and energy efficiency. However, considering DNN is typically computational-intensive and resource-hungry, conventional placement approaches ignore the influence of multi-dimensional resource requirements (processor, memory, etc.), which may degrade the real-time performance. Moreover, with the increasing scale of geo-distributed edge servers, centralized decision-making is still challenging to find the optimal strategies effectively. To overcome these shortcomings, in this paper we propose a game theoretic DNN placement approach in EI-enabled IoT. First, a DNN placement optimization problem is formulated to maximize system benefits, which is proven to be \(\mathcal {N}\mathcal {P}\)-hard and model the original problem as an exact potential game (EPG). Moreover, an EPG-based DNN model placement algorithm, named EPOL, is designed for edge servers to make sub-optimal strategies independently and theoretical analysis is possessed to guarantee the performance of EPOL. Finally, real-world dataset based experimental results corroborate the superiority and effectiveness of EPOL.
Bug fixing holds significant importance in software development and maintenance. Recent research has made substantial strides in exploring the potential of large language models (LLMs) for automatically resolving soft...
详细信息
Bug fixing holds significant importance in software development and maintenance. Recent research has made substantial strides in exploring the potential of large language models (LLMs) for automatically resolving software bugs. However, a noticeable gap in existing approaches lies in the oversight of collaborative facets intrinsic to bug resolution, treating the process as a single-stage endeavor. Moreover, most approaches solely take the buggy code snippet as input for LLMs during the patch generation stage. To mitigate the aforementioned limitations, we introduce a novel stage-wise framework named PATCH. Specifically, we first augment the buggy code snippet with corresponding dependence context and intent information to better guide LLMs in generating the correct candidate patches. Additionally, by taking inspiration from bug management practices, we decompose the bug-fixing task into four distinct stages: bug reporting, bug diagnosis, patch generation, and patch verification. These stages are performed interactively by LLMs, aiming to simulate the collaborative behavior of programmers during the resolution of software bugs. By harnessing these collective contributions, PATCH effectively enhances the bug-fixing capability of LLMs. We implement PATCH by employing the powerful dialogue-based LLM ChatGPT. Our evaluation on the widely used bug-fixing benchmark BFP demonstrates that PATCH has achieved better performance than state-of-the-art LLMs.
ICAT is the oldest international conference on virtual reality and tele-existence. ICAT 2006 not only looked for innovations in the technology itself, but also explored novel ways to transfer and express information a...
详细信息
ISBN:
(数字)9783540497790
ISBN:
(纸本)9783540497769
ICAT is the oldest international conference on virtual reality and tele-existence. ICAT 2006 not only looked for innovations in the technology itself, but also explored novel ways to transfer and express information and creative ideas to the society and people. The 16th International Conference on Artificial Reality and Telexistence was held at the Zhejiang University of Technology, Hangzhou, P. R. China from November 29 to December 1, 2006. The main purpose of the conference is to provide opportunities for researchers and practitioners to present their research findings and exchange opinions on the development and use of such systems. The conference included plenary invited talks, workshops, tutorials, and paper presentation tracks. The main conference received 523 submissions in total from 21 different countries, including China (mainland, Hong Kong, Taiwan), USA, UK, Germany, Austria, France, Australia, Canada, Korea, Japan, Malaysia, Mexico, etc. , of which 138 papers were accepted for this volume and 11 papers were invited to submit extended versions for a special issue of International Journal of Virtual Reality (IJVR, 5(4)).
This book constitutes the refereed proceedings of the 11th Annual Conference on Theory and Applications of Models of Computation, TAMC 2014, held in Chennai, India, in April 2014. The 27 revised full papers presented ...
详细信息
ISBN:
(数字)9783319060897
ISBN:
(纸本)9783319060880
This book constitutes the refereed proceedings of the 11th Annual Conference on Theory and Applications of Models of Computation, TAMC 2014, held in Chennai, India, in April 2014. The 27 revised full papers presented were carefully reviewed and selected from 112 submissions. The papers explore the algorithmic foundations, computational methods and computing devices to meet today's and tomorrow's challenges of complexity, scalability and sustainability, with wide-ranging impacts on everything from the design of biological systems to the understanding of economic markets and social networks.
This state-of-the-art survey offers a renewed and refreshing focus on the progress in evolutionary computation, in neural networks, and in fuzzy systems. The book presents the expertise and experiences of leading rese...
详细信息
ISBN:
(数字)9783642306877
ISBN:
(纸本)9783642306860
This state-of-the-art survey offers a renewed and refreshing focus on the progress in evolutionary computation, in neural networks, and in fuzzy systems. The book presents the expertise and experiences of leading researchers spanning a diverse spectrum of computational intelligence in these areas. The result is a balanced contribution to the research area of computational intelligence that should serve the community not only as a survey and a reference, but also as an inspiration for the future advancement of the state of the art of the field. The 13 selected chapters originate from lectures and presentations given at the IEEE World Congress on Computational Intelligence, WCCI 2012, held in Brisbane, Australia, in June 2012.
Carefully perturbing adversarial inputs degrades the performance of traditional machine learning (ML) models. Adversarial machine learning (AML) that takes adversaries into account during training and learning emerges...
详细信息
Carefully perturbing adversarial inputs degrades the performance of traditional machine learning (ML) models. Adversarial machine learning (AML) that takes adversaries into account during training and learning emerges as a valid technique to defend against attacks. Due to the complexity and uncertainty of adversaries’ attack strategies, researchers utilize game theory to study the interactions between an adversary and an ML system designer. By configuring different game rules and analyzing game outcomes in an adversarial game, it is possible to effectively predict attack strategies and to produce optimal defense strategies for the system designer. However, the literature still lacks a holistic review of adversarial games in AML. In this paper, we extend the scope of previous surveys and provide a thorough overview of existing game theoretical approaches in AML for adaptively defending against adversarial attacks. For evaluating these approaches, we propose a set of metrics to discuss their merits and drawbacks. Finally, based on our literature review and analysis, we raise several open problems and suggest interesting research directions worthy of special investigation.
暂无评论