Article
Computer Science, Information Systems
Jianping Dong, Gexiang Zhang, Biao Luo, Qiang Yang, Dequan Guo, Haina Rong, Ming Zhu, Kang Zhou
Summary: This paper proposes a distributed adaptive optimization spiking neural P system (DAOSNPS) that can solve combinatorial optimization problems without the help of evolutionary algorithms or swarm intelligence algorithms. Extensive experiments demonstrate its superiority over other methods.
INFORMATION SCIENCES
(2022)
Article
Computer Science, Information Systems
Jianping Dong, Gexiang Zhang, Biao Luo, Haina Rong
Summary: An extended numerical spiking neural (ENSN P) system is proposed to solve continuous constrained optimization problems. In ENSN P systems, production functions are selected by probability to achieve updated parameters. Experimental results show that the proposed method outperforms or is competitive with other 28 optimization algorithms in five benchmarks.
INFORMATION SCIENCES
(2023)
Article
Computer Science, Artificial Intelligence
Zeqiong Lv, Tingting Bao, Nan Zhou, Hong Peng, Xiangnian Huang, Agustin Riscos-Nunez, Mario J. Perez-Jimenez
Summary: The paper introduces a new variant of spiking neural P systems, called SNP-ECR systems, which are distributed parallel computing models with a stronger firing control mechanism. The new spiking rule ECR allows neurons to send different numbers of spikes, providing greater flexibility in computation. It is proven that SNP-ECR systems are Turing universal as number generating/accepting devices and can function as universal function computing devices.
INTERNATIONAL JOURNAL OF NEURAL SYSTEMS
(2021)
Article
Computer Science, Artificial Intelligence
Ming Zhu, Qiang Yang, Jianping Dong, Gexiang Zhang, Xiantai Gou, Haina Rong, Prithwineel Paul, Ferrante Neri
Summary: OSNPS is a membrane computing model that directly derives an approximate solution to combinatorial problems, specifically the 0/1 knapsack problem, using a family of parallel Spiking Neural P Systems (SNPS) and a Guider algorithm. However, its performance is only competitive with modern metaheuristics in low dimensional cases.
INTERNATIONAL JOURNAL OF NEURAL SYSTEMS
(2021)
Article
Computer Science, Information Systems
Zhang Sun, Luis Valencia -Cabrera, Guimin Ning, Xiaoxiao Song
Summary: Spiking neural P systems are an abstraction of the structure and function of nervous systems and neurons. SNP-WOD systems, a new class of these systems, remove the mechanism of duplication and allow for the amplification of pulses during the firing of spiking rules. These systems have computational properties and can generate numbers.
INFORMATION SCIENCES
(2022)
Article
Computer Science, Artificial Intelligence
Tingfang Wu, Qiang Lyu, Linqiang Pan
Summary: This study explores spiking neural P systems (SNP systems) and their variant evolution-communication SNP (ECSNP) systems, demonstrating the Turing universality of ECSNP systems as number-generating devices and highlighting the critical impact of the quantity of spikes in neurons on the computational power of ECSNP systems.
INTERNATIONAL JOURNAL OF NEURAL SYSTEMS
(2021)
Article
Computer Science, Information Systems
Xiaoxiao Song, Luis Valencia-Cabrera, Hong Peng, Jun Wang
Summary: This paper introduces a new neural computing model - spiking neural P systems with autapses (SNP-AU systems) and demonstrates their ability to generate Turing-computable numbers. By building an SNP-AU system with 53 neurons and providing a universal machine, the universality of its computing function is shown.
INFORMATION SCIENCES
(2021)
Article
Computer Science, Information Systems
Suxia Jiang, Yijun Liu, Bowen Xu, Junwei Sun, Yanfeng Wang
Summary: In this study, asynchronous numerical spiking neural (ANSN) P systems are investigated by combining set theory and threshold control strategy. It is proved that ANSN P systems are Turing universal and capable of processing information.
INFORMATION SCIENCES
(2022)
Review
Chemistry, Multidisciplinary
Yicen Liu, Ying Chen, Prithwineel Paul, Songhai Fan, Xiaomin Ma, Gexiang Zhang
Summary: This paper discusses the application of spiking neural P systems in fault diagnosis in power systems, and explores their efficiency in different power equipment systems as well as future research directions.
APPLIED SCIENCES-BASEL
(2021)
Article
Computer Science, Information Systems
Tingfang Wu, Luping Zhang, Qiang Lyu, Yu Jin
Summary: Asynchronous spiking neural P (AsynSN P) systems are distributed and parallel computational models inspired by biological neurons. This study introduces a control mechanism of local synchronization at the rule level and examines the computational power of different systems. The results show that local synchronization of rules can improve the computational capability of the systems.
INFORMATION SCIENCES
(2022)
Article
Engineering, Chemical
Aleksei Dominic C. Fernandez, Reyster M. Fresco, Francis George C. Cabarle, Ren Tristan A. de la Cruz, Ivan Cedric H. Macababayao, Korsie J. Ballesteros, Henry N. Adorna
Summary: SN P systems, inspired by spiking neurons, are effective in solving computationally hard problems and have been actively researched for both theory and application. Tools like Snapse aim to provide a visual and easy-to-use platform for simulating and creating SN P systems, thus accelerating research progress.
Article
Chemistry, Multidisciplinary
Xiong Chen, Ping Guo
Summary: This paper studies four basic arithmetic operations and improves the parallelization of addition and multiplication methods. It designs more effective SNPS for natural number addition, multiplication, subtraction, and division based on multiple subtractions. The proposed SNPS is verified to be effective through examples. Compared with similar SNPS, our system reduces the number of neurons used and the time overhead for addition operation by 50% and 33% respectively, and reduces the number of neurons used for multiplication operation by 40%.
APPLIED SCIENCES-BASEL
(2023)
Article
Computer Science, Artificial Intelligence
Resmi RamachandranPillai, Michael Arock
Summary: The paper introduces a new variant of vehicle routing problem that combines improved algorithm and neural systems, proposing spiking neural firefly optimization to solve dynamic VRP. By working in parallel across multiple neural systems, the proposed method has made significant progress in finding optimal solutions.
NEURAL COMPUTING & APPLICATIONS
(2021)
Article
Computer Science, Artificial Intelligence
Yuping Liu, Yuzhen Zhao
Summary: This work introduces the phenomenon of lateral inhibition in biological nervous systems into spiking neural P systems and proposes SN P systems with lateral inhibition (LISN P systems). By adjusting the lateral distance, lateral inhibition can affect the information transmission between neurons, and the designed LISN P systems can be used to generate arbitrary numbers and perform function computation, demonstrating computational completeness.
Article
Computer Science, Artificial Intelligence
Xiaoxiao Song, Luis Valencia-Cabrera, Hong Peng, Jun Wang, Mario J. Perez-Jimenez
Summary: This paper introduces SNP-DS systems as a powerful computing model based on the characteristics and communication of neurons, with the feature of neurons receiving spikes at different instants through delaying synapses. Additionally, two small universal SNP-DS systems are constructed to compute functions, with a simulator provided for experimental validation.
INTERNATIONAL JOURNAL OF NEURAL SYSTEMS
(2021)
Article
Computer Science, Artificial Intelligence
Jianping Dong, Biao Luo, Gexiang Zhang
Summary: In this study, an automatic design method of arithmetic operation SN P systems is proposed to achieve arithmetic operations by selecting redundant rules. Experimental results demonstrate the feasibility and effectiveness of this method.
Article
Computer Science, Artificial Intelligence
Yu Xue, Haokai Zhu, Ferrante Neri
Summary: This paper proposes a hybrid feature selection algorithm, MOFS-RFGA, which is based on a multi-objective algorithm with ReliefF. By combining the advantages of filter and wrapper methods, these two types of algorithms are hybridized to enhance the capability of solving feature selection problems. The experiments show that MOFS-RFGA can fully utilize the advantages of filter and wrapper methods, outperforming the comparison algorithms on multiple datasets while maintaining good classification performance.
APPLIED SOFT COMPUTING
(2023)
Article
Computer Science, Artificial Intelligence
Yu Xue, Yiling Tong, Ferrante Neri
Summary: This paper proposes a hybrid gradient descent search algorithm (HGDSA) for training fully-connected neural networks. The algorithm combines gradient descent strategies and BP to perform global and local searches. Experimental results demonstrate that HGDSA has powerful abilities in both global and local searches.
APPLIED INTELLIGENCE
(2023)
Article
Computer Science, Artificial Intelligence
Pengcheng Jiang, Yu Xue, Ferrante Neri
Summary: This paper proposes a multi-objective pruning method (MOP-FMS) based on feature map selection, which takes the number of FLOPs as a pruning objective in addition to the accuracy rate. The authors design an efficient search space, domain-specific crossover and mutation operators, decoding and pruning methods, and use multi-objective optimization for evaluation. Experimental results demonstrate that the proposed method achieves higher pruning rate without sacrificing the accuracy rate.
APPLIED SOFT COMPUTING
(2023)
Article
Computer Science, Artificial Intelligence
Xingyang Liu, Gexiang Zhang, Muhammad Shahid Mastoi, Ferrante Neri, Yang Pu
Summary: This article proposes a novel human-simulated fuzzy membrane control system for joint angle control to ensure the stable locomotion of biped robots. The control system incorporates a human-simulated intelligent control algorithm and a membrane architecture to improve control accuracy and real-time performance by simulating the human brain and membrane computing.
INTEGRATED COMPUTER-AIDED ENGINEERING
(2023)
Article
Computer Science, Artificial Intelligence
Yu Xue, Yixia Zhang, Ferrante Neri
Summary: This paper proposes a new method that combines Evolutionary Algorithms (EAs) and Attention Mechanisms to train GANs. The method progressively improves the activation of generator weights using an EA, resulting in better-performing configurations for image generation. Additionally, the use of a channel attention mechanism improves image quality and texture details. Experimental results show that the proposed method, called Attention evolutionary GAN (AevoGAN), alleviates the training instability problems of CycleGAN and produces higher quality images compared to existing methods in terms of IS, FID, and KID.
INTERNATIONAL JOURNAL OF NEURAL SYSTEMS
(2023)
Article
Automation & Control Systems
Pengcheng Jiang, Yu Xue, Ferrante Neri
Summary: Dropout is an effective method for training deep neural networks by deactivating some neurons to mitigate overfitting. This paper proposes a novel approach to guide the dropout rate using an evolutionary algorithm, allowing for more flexibility in training. Experimental results demonstrate that this method consistently outperforms other dropout methods, including state-of-the-art techniques.
ENGINEERING APPLICATIONS OF ARTIFICIAL INTELLIGENCE
(2023)
Article
Computer Science, Artificial Intelligence
Qian Liu, Lifan Long, Hong Peng, Jun Wang, Qian Yang, Xiaoxiao Song, Agustin Riscos-Nunez, Mario J. Perez-Jimenez
Summary: This article proposes a new variant of SNP systems, called GSNP systems, which are composed of gated neurons and introduce two gated mechanisms to control the updating of states in neurons. The GSNP model based on gated neurons is developed for time series prediction and is evaluated against benchmark models, demonstrating its availability and effectiveness.
IEEE TRANSACTIONS ON NEURAL NETWORKS AND LEARNING SYSTEMS
(2023)
Article
Chemistry, Multidisciplinary
Hua Yang, Shuxiang Zhang, Hao Shen, Gexiang Zhang, Xingquan Deng, Jianglin Xiong, Li Feng, Junxiong Wang, Haifeng Zhang, Shenyang Sheng
Summary: In this paper, a multi-layer feature fusion text classification model called CAC, based on the Combination of CNN and Attention, is proposed. The experimental results show that the CAC model outperforms models solely relying on Attention mechanism and other models based on CNN, RNN, and Attention in terms of accuracy and performance.
APPLIED SCIENCES-BASEL
(2023)
Article
Computer Science, Artificial Intelligence
Yu Xue, Changchang Lu, Ferrante Neri, Jiafeng Qin
Summary: This article proposes progressive partial channel connections based on channel attention for differentiable architecture search (PA-DARTS) to solve the instability and performance collapse problems in neural architecture search. Experimental results showed that PA-DARTS achieved 97.59% and 83.61% classification accuracy on CIFAR-10 and CIFAR-100, respectively, and 75.3% classification accuracy on ImageNet.
IEEE TRANSACTIONS ON EMERGING TOPICS IN COMPUTATIONAL INTELLIGENCE
(2023)
Article
Automation & Control Systems
Qian Liu, Hong Peng, Lifan Long, Jun Wang, Qian Yang, Mario J. Perez-Jimenez, David Orellana-Martin
Summary: SNP systems are neural-like computing models that are inspired by spiking neurons and have applications in chaotic time series forecasting. Nonlinear SNP systems with autapses (NSNP-AU systems) are proposed in this study, which have nonlinearity in spike consumption, generation, and gate functions. Based on NSNP-AU systems, a recurrent-type prediction model for chaotic time series, called the NSNP-AU model, is developed and implemented using a deep learning framework. Experimental results show the superiority of the NSNP-AU model in chaotic time series forecasting.
IEEE TRANSACTIONS ON CYBERNETICS
(2023)
Article
Computer Science, Information Systems
Jin Zhou, Kang Zhou, Gexiang Zhang, Ferrante Neri, Wangyang Shen, Weiping Jin
Summary: In this paper, the authors focus on the issue of multi-objective optimisation problems with redundant variables and indefinite objective functions (MOPRVIF) in practical problem-solving. They propose a dual data-driven method for solving this problem, which consists of eliminating redundant variables, constructing objective functions, selecting evolution operators, and using a multi-objective evolutionary algorithm. The experiments conducted on two different problem domains demonstrate the effectiveness, practicality, and scalability of the proposed method.
INFORMATION SCIENCES
(2024)
Article
Chemistry, Analytical
Qiang Yang, Song Ma, Gexiang Zhang, Kaiyi Xian, Lijia Zhang, Zhongyu Dai
Summary: In this paper, a new assistance navigation method is proposed to solve the problem of the inspection robot being affected by grass during the inspection task. By improving the network structure and fusing sensor information, the robot can safely cross the grass area and improve the inspection efficiency.
Article
Computer Science, Artificial Intelligence
Thiago Z. Miranda, Diorge B. Sardinha, Ferrante Neri, Marcio P. Basgalupp, Ricardo Cerri
Summary: This paper proposes a novel grammar-based multi-objective neuroevolutionary method for generating Fully Convolutional Networks. The method includes an efficient way to encode skip connections, facilitates the description of complex search spaces and the injection of domain knowledge, and incorporates multi-input layers and upsampling. The proposed method outperforms previous grammar evolution algorithms, achieving 90.5% accuracy on CIFAR-10.
APPLIED SOFT COMPUTING
(2023)
Article
Computer Science, Artificial Intelligence
Yu Xue, Chenyi Zhang, Ferrante Neri, Moncef Gabbouj, Yong Zhang
Summary: The study proposes a feature selection method called EAR-FS based on an attention mechanism and hybrid metaheuristic, which reduces the number of features in high-dimensional data while ensuring classification accuracy.
KNOWLEDGE-BASED SYSTEMS
(2023)