Special Issue on Theoretical analysis of deep learning
摘要截稿:
全文截稿: 2018-09-15
影响因子: 4.438
期刊难度:
CCF分类: C类
中科院JCR分区:
• 大类 : 计算机科学 - 2区
• 小类 : 计算机:人工智能 - 2区
Overview
Research on deep learning has made significant progress in both theoretical investigation and practical applications such as learning or teaching assistants for children, the elderly or people with deficiencies. They even show potential for individual adaptation for learning. Deep learning algorithms also have the potential to enhance learning via new models. Deep learning has been shown to engage the learner adaptively; to motivate the development of model, dynamics behavior, bifurcation analysis, control et al. in the fields of memristor, neural networks, industrial network, learning system, intelligent algorithm, smart grid, robotic system and so on. However, this ever-changing world brings about new challenges in practical applications of neural information processing. Data sets from practical application growing rapidly are no long structured which causes traditional data processing approaches inadequate to deal with them. Moreover, due to the large volume and the scalability of ever-increasing data sets, it is becoming more difficult for traditional methods to keep up in real-time or near real-time for time-limited tasks. Indeed, for many cases, data is contaminated by noisy which leads to unreliable information. it is a challenge research topic to investigate the theories of deep learning. Deep learning of practical applications in big data analytics, Internet of thing (IoT) and cyber security becomes meaningful information when it is able to uncover unknown pattern and produce doable business insights.
This special issue is devoted to the Theoretical analysis of deep learning. It is aiming to publish the frontier of theories of deep learning in complexity environment and society. The issue will concentrate on presenting several theoretical and practical problems related to deep learning, and new discoveries and innovative ideas and improvements made in this field. With this special issue we aim at collecting an overview on theoretical state-of-the-art research contributions on deep learning.
The list of possible topics includes, but is not limited to:
- Mathematical analysis of deep neural networks
- Model design of general deep networks
- Deep neural network based control method
- Neurodynamics and its application
- Dynamic analysis of deep neural networks, i.e., bifurcation and chaos analysis
- New model of memristor-based system and its application
- Deep neural network based algorithms in smart grids
- Deep neural networks for image processing
- Novel deep network architecture for emerging nano-devices
- Efficient training analysis for deep learning
- Plug-in Electric Vehicle (PEV) management via learning systems
- New meta-heuristic algorithm and its application