A5101908375- Neural Networks and Applications
- Neural Networks and Reservoir Computing
- Machine Learning and ELM
- Advanced Memory and Neural Computing
- Blind Source Separation Techniques
- Time Series Analysis and Forecasting
- Chaos control and synchronization
- Energy Load and Power Forecasting
- Conducting polymers and applications
- Neural Networks Stability and Synchronization
- Magnetic properties of thin films
- Nonlinear Dynamics and Pattern Formation
- Neural dynamics and brain function
- Supercapacitor Materials and Fabrication
- Metaheuristic Optimization Algorithms Research
- Grey System Theory Applications
- Optical Network Technologies
- Stock Market Forecasting Methods
- Machine Learning in Materials Science
- Brain Tumor Detection and Classification
- Advanced Adaptive Filtering Techniques
- IoT Networks and Protocols
- Electromagnetic wave absorption materials
- Supramolecular Chemistry and Complexes
- Advanced Photocatalysis Techniques
Northeastern University
2020-2024
The Synergetic Innovation Center for Advanced Materials
2024
Nanjing Tech University
2010-2024
University of International Business and Economics
2023-2024
Nanjing Drum Tower Hospital
2024
Jilin Agricultural University
2024
Dalian University
2013-2023
Dalian University of Technology
2012-2023
Jiaxing University
2019-2020
Hebei Finance University
2019
The broad learning system (BLS) is an emerging approach for effective and efficient modeling of complex systems. inputs are transferred placed in the feature nodes, then sent into enhancement nodes nonlinear transformation. structure a BLS can be extended wide sense. Incremental algorithms designed fast expansion. Based on typical BLSs, novel recurrent (RBLS) proposed this paper. units recurrently connected, purpose capturing dynamic characteristics time series. A sparse autoencoder used to...
High-dimensional and large-scale time series processing has aroused considerable research interests during decades. It is difficult for traditional methods to reveal the evolution state in dynamical systems discover relationship among variables automatically. In this paper, we propose a unified framework nonuniform embedding, system revealing, prediction, termed as Structured Manifold Broad Learning System (SM-BLS). The structured manifold learning introduced embedding unsupervised...
Echo state network (ESN) is a new kind of recurrent neural with randomly generated reservoir structure and an adaptable linear readout layer. It has been widely employed in the field time series prediction. However, when high-dimensional reservoirs are utilized to predict multivariate series, there may be collinearity problem. In this paper, overcome problem obtain sparse solution, we propose model-adaptive elastic ESN, which adaptive net algorithm used calculate unknown weights. combines...
Echo state network is a novel kind of recurrent neural networks, with trainable linear readout layer and large fixed connected hidden layer, which can be used to map the rich dynamics complex real-world data sets. It has been extensively studied in time series prediction. However, there may an ill-posed problem caused by number training samples less than size layer. In this brief, Laplacian echo (LAESN), proposed overcome obtain low-dimensional output weights. First, multivariate into...
Kernel recursive least squares (KRLS) is a kind of kernel methods, which has attracted wide attention in the research time series online prediction. It low computational complexity and updates form. However, as data size increases, calculating inverse matrix will raise. And it some difficulties accommodating time-varying environments. Therefore, we have presented an improved KRLS algorithm for multivariate chaotic Approximate linear dependency, dynamic adjustment, coherence criterion are...
With the applications of Internet Things (IoT) for smart cities, real-time performance a large number network packets is facing serious challenge. Thus, how to improve emergency response has become critical issue. However, traditional packet scheduling algorithms cannot meet requirements large-scale IoT system cities. To address this shortcoming, paper proposes EARS, an efficient data-emergency-aware scheme EARS describes information with priority and deadline. Each source node informs...
Accurate prediction of solar irradiance is beneficial in reducing energy waste associated with photovoltaic power plants, preventing system damage caused by the severe fluctuation irradiance, and stationarizing output integration between different grids. Considering randomness multiple dimension weather data, a hybrid deep learning model that combines gated recurrent unit (GRU) neural network an attention mechanism proposed forecasting changes four seasons. In first step, Inception ResNet...
Multivariate chaotic time series prediction is a hot research topic, the goal of which to predict future based on past observations. Echo state networks (ESNs) have recently been widely used in prediction, but there may be an ill-posed problem for large number unknown output weights. To solve this problem, we propose hybrid regularized ESN, employs sparse regression with L1/2 regularization and L2 compute The penalty shows many attractive properties, such as unbiasedness sparsity. presents...
In multivariate chaotic time series prediction, correlation analysis is important for reducing input dimensions and improving prediction performance. Grey relational (GRA) has proved to be an effective method data analysis, especially inexact incomplete data. GRA, points are usually regarded as objects, the distance between or concave convex degree mostly used measure correlations. However, with discrete variables, results always tend have some deviations when using prior GRA methods....
State space reconstruction is the foundation of chaotic system modeling. Selection reconstructed variables essential to analysis and prediction multivariate time series. As most existing state theorems deal with univariate series, we have presented a novel nonuniform method using information criterion for We derived new based on low dimensional approximation joint mutual delay selection, which can be solved efficiently through use an intelligent optimization algorithm computation complexity....
ABSTRACT In this study, a method is developed to fabricate sulfonated poly (ether ether ketone)/phosphotungstic acid‐polyaniline (SPEEK/HPW‐PANI) membranes by in situ polymerization of aniline for the purpose decreasing weight loss HPW membranes. The synthesis involves production SPEEK/HPW hybrid membrane followed different layer PANI coatings on surface, and subsequent treatment using drying vacuum procedures. scanning electronic microscopy images showed that had good compatibility with...
Spatio-temporal series prediction has attracted increasing attention in the field of meteorology recent years. The spatial and temporal joint effect makes predictions challenging. Most existing spatio-temporal models are computationally complicated. To develop an accurate but easy-to-implement model, this paper designs a novel model based on echo state networks. For real-world observed meteorological data with randomness large changes, we use cubic spline method to bridge gaps between...
Ground surface settlement forecasting in the process of tunnel construction is one most important techniques towards sustainable city development and preventing serious damages, such as landscape collapse. It evident that modern artificial intelligence (AI) models, neural network, extreme learning machine, support vector regression, are capable providing reliable results for settlement. However, two limitations exist current techniques. First, data provided by company usually univariate...