Deep embedded self-organizing maps for joint representation learning and topology-preserving clustering

2021 ◽  
Author(s):  
Florent Forest ◽  
Mustapha Lebbah ◽  
Hanene Azzag ◽  
Jérôme Lacaille
Keyword(s):  
Representation Learning ◽  
Self Organizing Maps ◽  
And Topology ◽  
Joint Representation ◽  
Self Organizing

scholarly journals Self-Organizing Representation Learning

2021 ◽  
Author(s):  
noureddine kermiche
Keyword(s):  
Data Augmentation ◽  
Representation Learning ◽  
Self Organization ◽  
Learning Methods ◽  
Self Organizing Maps ◽  
Output Layer ◽  
Data Representations ◽  
Augmentation Techniques ◽  
Using Data ◽  
Self Organizing

Using data augmentation techniques, unsupervised representation learning methods extract features from data by training artificial neural networks to recognize that different views of an object are just different instances of the same object. We extend current unsupervised representation learning methods to networks that can self-organize data representations into two-dimensional (2D) maps. The proposed method combines ideas from Kohonen’s original self-organizing maps (SOM) and recent development in unsupervised representation learning. A ResNet backbone with an added 2D <i>Softmax</i> output layer is used to organize the data representations. A new loss function with linear complexity is proposed to enforce SOM requirements of winner-take-all (WTA) and competition between neurons while explicitly avoiding collapse into trivial solutions. We show that enforcing SOM topological neighborhood requirement can be achieved by a fixed radial convolution at the 2D output layer without having to resort to actual radial activation functions which prevented the original SOM algorithm from being extended to nowadays neural network architectures. We demonstrate that when combined with data augmentation techniques, self-organization is a simple emergent property of the 2D output layer because of neighborhood recruitment combined with WTA competition between neurons. The proposed methodology is demonstrated on SVHN and CIFAR10 data sets. The proposed algorithm is the first end-to-end unsupervised learning method that combines data self-organization and visualization as integral parts of unsupervised representation learning.

scholarly journals Self-Organizing Representation Learning

2021 ◽  
Author(s):  
noureddine kermiche
Keyword(s):  
Data Augmentation ◽  
Representation Learning ◽  
Self Organization ◽  
Learning Methods ◽  
Self Organizing Maps ◽  
Output Layer ◽  
Data Representations ◽  
Augmentation Techniques ◽  
Using Data ◽  
Self Organizing

Using data augmentation techniques, unsupervised representation learning methods extract features from data by training artificial neural networks to recognize that different views of an object are just different instances of the same object. We extend current unsupervised representation learning methods to networks that can self-organize data representations into two-dimensional (2D) maps. The proposed method combines ideas from Kohonen’s original self-organizing maps (SOM) and recent development in unsupervised representation learning. A ResNet backbone with an added 2D <i>Softmax</i> output layer is used to organize the data representations. A new loss function with linear complexity is proposed to enforce SOM requirements of winner-take-all (WTA) and competition between neurons while explicitly avoiding collapse into trivial solutions. We show that enforcing SOM topological neighborhood requirement can be achieved by a fixed radial convolution at the 2D output layer without having to resort to actual radial activation functions which prevented the original SOM algorithm from being extended to nowadays neural network architectures. We demonstrate that when combined with data augmentation techniques, self-organization is a simple emergent property of the 2D output layer because of neighborhood recruitment combined with WTA competition between neurons. The proposed methodology is demonstrated on SVHN and CIFAR10 data sets. The proposed algorithm is the first end-to-end unsupervised learning method that combines data self-organization and visualization as integral parts of unsupervised representation learning.

A Psychographic Segmentation of Kuwaiti Travelers Using Self-Organizing Maps

2019 ◽  
Vol 24 (1) ◽  
pp. 87-92 ◽  
Author(s):  
Yvette Reisinger ◽  
Mohamed M. Mostafa ◽  
John P. Hayes
Keyword(s):  
Self Organizing Maps ◽  
Self Organizing

Self Organizing Maps, Pattern Recognition and Financial Crises

2011 ◽  
Author(s):  
Sylvain Barthelemy ◽  
Pascal Devaux ◽  
Francois Faure ◽  
Matthieu Pautonnier
Keyword(s):  
Pattern Recognition ◽  
Financial Crises ◽  
Self Organizing Maps ◽  
Self Organizing

Using self organizing maps to analyze larval fish assemblage vertical dynamics through environmental-ontogenetic gradients

2021 ◽  
pp. 107410
Author(s):  
I. Álvarez ◽  
J.S. Font-Muñoz ◽  
I. Hernández-Carrasco ◽  
C. Díaz-Gil ◽  
P.M. Salgado-Hernanz ◽  
...  
Keyword(s):  
Fish Assemblage ◽  
Larval Fish ◽  
Self Organizing Maps ◽  
Self Organizing

scholarly journals The Spread of the COVID-19 Outbreak in Brazil: An Overview by Kohonen Self-Organizing Map Networks

Medicina ◽  
2021 ◽  
Vol 57 (3) ◽  
pp. 235
Author(s):  
Diego Galvan ◽  
Luciane Effting ◽  
Hágata Cremasco ◽  
Carlos Adam Conte-Junior
Keyword(s):  
The Self ◽  
Self Organizing Map ◽  
Virus Spread ◽  
Self Organizing Maps ◽  
The North ◽  
Som Clustering ◽  
Unsupervised Neural Networks ◽  
Mining Tools ◽  
Temporal Spread ◽  
Self Organizing

Background and objective: In the current pandemic scenario, data mining tools are fundamental to evaluate the measures adopted to contain the spread of COVID-19. In this study, unsupervised neural networks of the Self-Organizing Maps (SOM) type were used to assess the spatial and temporal spread of COVID-19 in Brazil, according to the number of cases and deaths in regions, states, and cities. Materials and methods: The SOM applied in this context does not evaluate which measures applied have helped contain the spread of the disease, but these datasets represent the repercussions of the country’s measures, which were implemented to contain the virus’ spread. Results: This approach demonstrated that the spread of the disease in Brazil does not have a standard behavior, changing according to the region, state, or city. The analyses showed that cities and states in the north and northeast regions of the country were the most affected by the disease, with the highest number of cases and deaths registered per 100,000 inhabitants. Conclusions: The SOM clustering was able to spatially group cities, states, and regions according to their coronavirus cases, with similar behavior. Thus, it is possible to benefit from the use of similar strategies to deal with the virus’ spread in these cities, states, and regions.

scholarly journals Geospatial Analysis of Extreme Weather Events in Nigeria (1985–2015) Using Self-Organizing Maps

2017 ◽  
Vol 2017 ◽  
pp. 1-11 ◽  
Author(s):  
Adeoluwa Akande ◽  
Ana Cristina Costa ◽  
Jorge Mateu ◽  
Roberto Henriques
Keyword(s):  
Extreme Precipitation ◽  
Large Scale ◽  
Northern Region ◽  
Extreme Weather Events ◽  
Self Organizing Map ◽  
Spatial And Temporal Patterns ◽  
Self Organizing Maps ◽  
The North ◽  
Precipitation Regimes ◽  
Self Organizing

The explosion of data in the information age has provided an opportunity to explore the possibility of characterizing the climate patterns using data mining techniques. Nigeria has a unique tropical climate with two precipitation regimes: low precipitation in the north leading to aridity and desertification and high precipitation in parts of the southwest and southeast leading to large scale flooding. In this research, four indices have been used to characterize the intensity, frequency, and amount of rainfall over Nigeria. A type of Artificial Neural Network called the self-organizing map has been used to reduce the multiplicity of dimensions and produce four unique zones characterizing extreme precipitation conditions in Nigeria. This approach allowed for the assessment of spatial and temporal patterns in extreme precipitation in the last three decades. Precipitation properties in each cluster are discussed. The cluster closest to the Atlantic has high values of precipitation intensity, frequency, and duration, whereas the cluster closest to the Sahara Desert has low values. A significant increasing trend has been observed in the frequency of rainy days at the center of the northern region of Nigeria.

Sign in / Sign up

Export Citation Format

Share Document