A5100954943- Machine Learning and ELM
- Advanced Decision-Making Techniques
- Neural Networks and Applications
- Advanced Sensor and Control Systems
- Face and Expression Recognition
- Advanced machining processes and optimization
- Advanced Computational Techniques and Applications
- Obstructive Sleep Apnea Research
- Sleep and related disorders
- Phase Change Materials Research
- Domain Adaptation and Few-Shot Learning
- Thermodynamic and Exergetic Analyses of Power and Cooling Systems
- Sleep and Wakefulness Research
- Music Technology and Sound Studies
- Extenics and Innovation Methods
- Collaboration in agile enterprises
- Advanced Memory and Neural Computing
- Mathematical Dynamics and Fractals
- Advanced Thermodynamics and Statistical Mechanics
- Organic Light-Emitting Diodes Research
- Geothermal Energy Systems and Applications
- Music and Audio Processing
- Image Processing and 3D Reconstruction
- Luminescence and Fluorescent Materials
- Enzyme Production and Characterization
Hebei University of Technology
2023-2024
Second Affiliated Hospital of Zhejiang University
2023
Jilin Engineering Normal University
2017-2020
Xinxiang Medical University
2010-2015
Southwest University
2014-2015
Institute of Scientific and Technical Information
2002
Prediction of library borrowing flow plays an important role in controlling the quality book collection and cataloging work library. Traditional time series prediction methods are hard to model because it is a nonlinear dynamical process has nonstationary stochastic character. Based on support vector machine theory chaotic prediction, new method proposed predict flow. The experiments show this reasonable solve problem certain value both practice.
Recently, extreme learning machines (ELMs) have been a promising tool in solving wide range of regression and classification applications. However, when modeling multiple related tasks which only limited training data per task are available the dimension is low, ELMs generally hard to get impressive performance due little help from informative domain knowledge across tasks. To solve this problem, paper extends ELM scenario multi-task (MTL). First, based on assumption that model parameters...
Early mechanical failure of CNC milling motorized spindle having a hidden and complexity is difficult to quickly accurately identify early machines milling.In this paper, latent subtle abnormal vibration signals detected by PeakVue.Fuzzy neural network diagnosis for each partial signal, then the sub-diagnosis as evidence, use D-S evidence theory global final diagnosis, further improve fault recognition rate.Using method rough set data mining obtain processing parts surface roughness...
The initial investment for a buried pipe system using soil as seasonal thermal storage medium is costly and not suitable small rural buildings. To promote the cost-effective heating in areas, this study proposes novel utilizing water with copper rods to enhance transfer between tank surrounding soil. An experimental platform was established discuss temperature variation characteristics investigate performance under different operating modes parameters. results show that of without linearly...
As the rapid increase of music data, Music Information Retrieval (MIR) have been receiving increasing attention in both academic and commercial spheres. Feature extraction is a crucial part many tasks. In recent years, deep learning approaches gained significant interest as way higher abstract representation from unlabeled data. this paper, we present system that can automatically extract relevant symbolic Firstly, The lower level features are extracted by using toolbox Music21, feature then...
The increasing heating demand calls for a correspondingly augmented capacity of the primary network. However, due to restriction existing pipes, can be finitely increased. Therefore, based on organic Rankine cycle (ORC) and vapor compression (VCC), novel thermal station is proposed realize self-augmented capacity. Based laws thermodynamics, mathematical model constructed, dynamic performances power generation are analyzed. Results show that supply water temperature network greatly influences...
Accurate and fast prediction of nonstationary time series is challenging great interest in both practical academic areas. In this paper, an online sequential extreme learning machine with new weighted strategy proposed for prediction. First, a leave-one-out (LOO) cross-validation error estimation data based on inversion block matrix. Second, the LOO proposed. This ranks samples’ importance by means each added sample then assigns various weights. Performance comparisons method other existing...
Aiming at the difficulty of fuzzy synthetic discrimination model NC machine, this paper proposes primary diagnosis results which change evaluation factors from high dimensional features to low one based on initial classifier.It reduces complexity and constructs a single stage for fault machine information fusion successfully.It puts forward method evaluating classifier classification ability tow aspects validity accuracy, whole weight allocation function according entropy, finally influence...
Coupled dynamical system is made up of several interactive dynamic systems, the coupling strengths which are decided by parameters. Recent researches have shown some powerful capability this kind in engineering fields. Gambling games usually considered as result-stochastic activities. In paper, we use discrete sampling time series attractor Lorenz chaotic to model a simple hammer-forfex-cloth gambling game. A coupled set game between "man" and "machine". Through great deal trial error, find...