Comparative Study of Fuzzy Methods for Response Integration in Ensemble Neural Networks for Pattern Recognition

Artificial neural networks and support vector machine (SVM), as two important tools, have widely applied in artificial intelligence and pattern recognition. In this paper, a comparative study has been done for making an analysis on their performances, when they are used in pattern recognition. Through theoretical analysis and confirmed by experimental results, a conclusion can be drawn that support vector machines have obvious advantages over those of traditional neural networks.

Download Full-text

Discrimination of Two-Mixture Fragrances Odor Using Artificial Odor Recognition System with Ensemble Backpropagation Neural Networks

Applied Mechanics and Materials ◽

10.4028/www.scientific.net/amm.303-306.1514 ◽

2013 ◽

Vol 303-306 ◽

pp. 1514-1518

Author(s):

Benyamin Kusumoputro ◽

Li Na

Keyword(s):

Neural Networks ◽

Pattern Recognition ◽

Recognition Rate ◽

Recognition System ◽

Quartz Resonator ◽

Learning Mechanisms ◽

Odor Recognition ◽

Pattern Recognition System ◽

Backpropagation Neural Networks ◽

Ensemble Neural Networks

The human sensory test is often used to obtain the sensory quantities of odors, however, the fluctuation of results due to the experts condition can cause discrepancies among panelists. We have developed an artificial odor recognition system using a quartz resonator sensor and backpropagation neural networks as the pattern recognition system in order to eliminate the disadvantage of human panelist system. The backpropagation neural networks shows high recognition rate for single component odor, however, become very low when it is used to discriminate mixture fragrances odor. In this paper we have proposed an ensemble of backpropagation neural networks as the pattern recognition system, and by using the ensemble learning mechanisms, the recognition rate is significantly increased, especially when using ensemble neural networks with five components.

Download Full-text

Review of Pattern Recognition by Self-Organizing Neural Networks.

Contemporary Psychology ◽

10.1037/004153 ◽

1995 ◽

Vol 40 (11) ◽

pp. 1110-1110

Author(s):

Stephen James Thomas

Keyword(s):

Neural Networks ◽

Pattern Recognition ◽

Self Organizing

Download Full-text

Levenshtein Augmentation Improves Performance of SMILES Based Deep-Learning Synthesis Prediction

10.26434/chemrxiv.12562121 ◽

2020 ◽

Author(s):

Dean Sumner ◽

Jiazhen He ◽

Amol Thakkar ◽

Ola Engkvist ◽

Esben Jannik Bjerrum

Keyword(s):

Neural Networks ◽

Pattern Recognition ◽

Deep Learning ◽

Recurrent Neural Networks ◽

Data Augmentation ◽

State Of The Art ◽

Sequence Similarity ◽

Learning Models ◽

Underlying Network

<p>SMILES randomization, a form of data augmentation, has previously been shown to increase the performance of deep learning models compared to non-augmented baselines. Here, we propose a novel data augmentation method we call “Levenshtein augmentation” which considers local SMILES sub-sequence similarity between reactants and their respective products when creating training pairs. The performance of Levenshtein augmentation was tested using two state of the art models - transformer and sequence-to-sequence based recurrent neural networks with attention. Levenshtein augmentation demonstrated an increase performance over non-augmented, and conventionally SMILES randomization augmented data when used for training of baseline models. Furthermore, Levenshtein augmentation seemingly results in what we define as <i>attentional gain </i>– an enhancement in the pattern recognition capabilities of the underlying network to molecular motifs.</p>

Download Full-text

Methods and tools of ensuring the operational reliability of complex industrial facilities.

Journal of Electrical and power engineering ◽

10.31474/2074-2630-2020-1-34-42 ◽

2020 ◽

Vol 14 (1) ◽

pp. 34-42

Author(s):

A. VAZHYNSKYI ◽

◽

S. ZHUKOV ◽

Keyword(s):

Neural Networks ◽

Pattern Recognition ◽

Artificial Neural Networks ◽

Expert Systems ◽

Residual Life ◽

Full Potential ◽

Scientific Publications ◽

Specific Element ◽

Industrial Facilities ◽

Artificial Neural

Approaches and algorithms for processing experimental data and data obtained as a result of using modern means of measuring equipment, selecting diagnostic parameters, pattern recognition, which constitute the methodological basis for developing methods and designing tools for creating a service system for complex industrial facilities based on predicting their performance and residual life are described in submitted article. Along with classical methods, methods based on using the full potential of the modern elemental base of microprocessor technology and the use of artificial neural networks, machine learning, and "big data" are discovered. The given examples can serve as the basis for constructing a methodology for the application of the considered approaches for organizing predictive maintenance of complex industrial equipment. An analytical review of a number of scientific publications showed that the creation of new automated diagnostic systems that can increase fault tolerance and extend the life of sophisticated modern power equipment is extremely relevant. For this, various approaches are applied, based on mathematical models, expert systems, artificial neural networks and other algorithms. Summarizing the results of scientific publications, it can be argued that the implementation of a systematic approach to the organization of repair service at the enterprise requires a comprehensive solution to the following urgent problems: • monitoring is formulated as the task of interrogating sensors and collecting information necessary for further analysis; • diagnostics, it is solved as tasks of identifying informative signs with further detection and classification of failures and anomalies in data sets; • improving the accuracy of algorithms aimed at pattern recognition; • condition forecasting is the task of assessing the current and accumulated readings of monitoring systems for making decisions regarding either a specific element of the complex or the facilities. Thus, modern technology make it possible to arrange arbitrarily complex algorithms. However, to use the full potential that artificial neural networks, expert systems, and classical methods for identifying and diagnosing equipment it is necessary to have a conceptual development of the foundations of building systems for organizing maintenance and repair of complex energy equipment

Download Full-text