An Overview of SOM Literature

Kohonen, Teuvo

doi:10.1007/978-3-642-56927-2_10

Cited by 11 publications

(8 citation statements)

References 315 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…It effectively captures non‐linear features as it does not assume stationarity or orthogonality, producing more realistic results compared to other clustering techniques (Gervais et al., 2020; Hewitson & Crane, 2002; Van Hulle, 2012). A detailed description of the SOM method can be found in Kohonen (2001).…”

Section: Methodsmentioning

confidence: 99%

Connecting Large‐Scale Meteorological Patterns to Extratropical Cyclones in CMIP6 Climate Models Using Self‐Organizing Maps

Gore,

Zarzycki,

Gervais

2023

Earth's Future

View full text Add to dashboard Cite

Extratropical cyclones (ETCs) are responsible for the majority of cool‐season extreme events in the northeastern United States (NEUS), often leading to high‐impact weather conditions that can have wide‐ranging socioeconomic impacts. Evaluating the ability of climate models to adequately simulate ETC dynamics is essential for improving model performance and increasing confidence in future projections used by stakeholders and policymakers. ETCs are traditionally studied using techniques such as case studies and synoptic typing, however, these approaches can be time‐consuming, require subjective analysis, and do not necessarily identify the coincident large‐scale meteorological patterns (LSMPs). Here, we apply self‐organizing maps (SOMs) as an automated machine‐learning approach to characterize the LSMPs and associated frequency and intensity of discrete ETC events over NEUS. The dominant patterns of geopotential height variability are identified through SOM analysis of five reanalysis products during the last four decades. ETC events are tracked using TempestExtremes and are integrated with SOMs to classify the accumulated cyclone activity (ACA) associated with each pattern. We then evaluate the skill of CMIP6 historical experiments in simulating the LSMPs and ETC events identified in the SOM. Our results identify a robust bias toward more zonal patterns, with models struggling to reproduce the more amplified patterns typically associated with the highest cyclone activity. While model resolution has some impact on simulation credibility, model configuration appears to be more important in LSMP representation. The vast majority of CMIP6 models produce too few ETCs, although model errors are distributed around historical reanalyses when ACA is normalized by storm frequency.

show abstract

Section: Methodsmentioning

confidence: 99%

Connecting Large‐Scale Meteorological Patterns to Extratropical Cyclones in CMIP6 Climate Models Using Self‐Organizing Maps

Gore,

Zarzycki,

Gervais

2023

Earth's Future

View full text Add to dashboard Cite

show abstract

“…Although some other techniques of examining multidimensional data exist (such as deep learning theory [14]), it was found that Kohonen’s Self-Organizing Map (SOM) would serve as the best tool for identifying groups of cities similar to each other while satisfying the mentioned requirements [15]. It enables the unsupervised learning process of a network and the mapping of multidimensional data and dependencies onto a smaller number of dimensions.…”

Section: Describing the Citiesmentioning

confidence: 99%

“…The aim is to group the towns, according to their similarities, on an array of nodes ( neurons ). This process, called mapping , was described by Kohonen [15] as an SOM algorithm. It converts complex relationships between high-dimensional data into simpler representations on one or two-dimensional arrays.…”

Section: Grouping the Citiesmentioning

confidence: 99%

“…Neurons in Kohonen’s maps (based on the definition [15]) have a fixed neighborhood, what is very important in projecting multidimensional similarities into (a simple) perception map.…”

Section: Grouping the Citiesmentioning

confidence: 99%

“…The authors chose the rectangular array as there are no crucial differences between two main kinds of topologies of a SOM (hexagonal and rectangular) (see [15], pages 109–110 and 159–161) in quality of preserving original multidimensional relationships among data.…”

Section: Grouping the Citiesmentioning

confidence: 99%

See 2 more Smart Citations

Scaling Laws in City Growth: Setting Limitations with Self-Organizing Maps

Cebrat

Sobczyński

2016

PLoS ONE

View full text Add to dashboard Cite

Do scaling relations always provide the means to anticipate the relationships between the size of cities, costs of maintenance, and the socio-economic benefits resulting from their growth? Scaling laws are considered a universal principle that describes the development of complex systems such as cities. It seems that regardless of their location or history, the growth of cities is associated with the super-linear or sublinear scaling of features such as the amount of space required, infrastructure, or human activities. However, the results of our research, based on grouping by Self-Organizing Maps, reveal some limitations in the application of scaling laws: the trends of urban growth behave in a different manner when we consider both a large and diverse collection of cities and a subset of cities alike. This finding complements the existing body of knowledge on the growth of cities and allows for a more accurate prediction of their future.

show abstract

Stress test of a biological early warning system with zebrafish (Danio rerio)

et al. 2016

View full text Add to dashboard Cite

The aim of this work was to develop a novel methodology to stress test the diagnostic capability of a video tracking system with zebrafish (Danio rerio), against two pre-established disturbances. Eight different treatments were tested varying the presence or absence of a toxicant (NaOCl) and two disturbances: the passing of a shadow (mimicking a predator) and entrapment of the fish. The concentration tested corresponded to a sublethal (1 % 24 h-LC) and short term exposure (2 h). A total of 56 organisms were tested resulting in 112 diagnoses (before and after the contamination). A statistical model of diagnosis was developed using Self-organizing Map (SOM) and Correspondence Analysis (CA). Sensitivity, specificity, accuracy, false positive and false negative values were calculated to evaluate the diagnostic performance. The disturbances did not negatively affect the capability of the model. In the presence of at least one of these variables, the diagnostic performance was similar or even superior to the baseline results without disturbances. Furthermore, the system produced a large number of correct diagnoses, at an ecologically relevant concentration of exposure, in a non-invasive way.

show abstract

An Overview of SOM Literature

Cited by 11 publications

References 315 publications

Connecting Large‐Scale Meteorological Patterns to Extratropical Cyclones in CMIP6 Climate Models Using Self‐Organizing Maps

Connecting Large‐Scale Meteorological Patterns to Extratropical Cyclones in CMIP6 Climate Models Using Self‐Organizing Maps

Scaling Laws in City Growth: Setting Limitations with Self-Organizing Maps

Stress test of a biological early warning system with zebrafish (Danio rerio)

Contact Info

Product

Resources

About