Improving the effectiveness of self-organizing map networks using a circular Kohonen layer

Kiang, Melody Y.; Kulkarni, Uday R.; Goul, Michael; Philippakis, Andreas S.; Chi, R.T.; Turban, Efraim

doi:10.1109/hicss.1997.663212

Cited by 9 publications

(7 citation statements)

References 22 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Weights are updated for winner node j* and all nodes in the neighborhood defined by NE 1 . (t) as shown in Fig 2. New weights as per proposed WNSOM are given by (3) and exponentially varying function is (4) where the range of weighted neighborhood factor is These equations are used to give decreasing update as we move away from winning cell, to cells in the neighborhood. This is obvious because the winning neuron should get the maximum reward as compared to others in the neighborhood.…”

Section: Proposed Wnsom Algorithmmentioning

confidence: 99%

“…Lo and Bavarian [3 J addressed the effect of neighborhood function selection on tne rate of convergence of the SOM algorithm. Kiang et al [ 4] developed a "circular" training algorithm that tries to overcome some of the ineffective topological representations caused by the "boundary" effect. Fritzke [5] proposed a new selforganizing neural network model that can determine shape as well as size of the network during the simulation in an incremental fashion.…”

mentioning

confidence: 99%

See 1 more Smart Citation

Devanagari Handwritten Digits Recognition using Weighted Neighborhood Self-Organizing Map

Kulkarni

Bhoyar

2002

IETE Journal of Research

View full text Add to dashboard Cite

Section: Proposed Wnsom Algorithmmentioning

confidence: 99%

mentioning

confidence: 99%

Devanagari Handwritten Digits Recognition using Weighted Neighborhood Self-Organizing Map

Kulkarni

Bhoyar

2002

IETE Journal of Research

View full text Add to dashboard Cite

“…Lo (1991) focused on the selection of neighborhood function, and Kiang (2006) proposed a circular training algorithm to overcome the "boundary" effect on topological representations. In another study, an incremental learning algorithm was applied (Jun et al, 1993).…”

Section: Introductionmentioning

confidence: 99%

Improving self‐organising information maps as navigational tools: a semantic approach

Lin

Brusilovsky²,

He³

2011

Online Information Review

View full text Add to dashboard Cite

• PurposeThe goal of the research is to explore whether the use of higher-level semantic features can help us build better SOM representation as measured from a human-centered perspective. We also explore an automatic evaluation method that utilizes human expert knowledge encapsulated in the structure of traditional textbooks to determine map representation quality.• Design/methodology/approach Two types of document representations involving semantic features have been explored: 1) using only one individual semantic feature, and 2) combining a semantic feature with keywords. A set of experiments were conducted to investigate the impact of semantic representation quality on the map. The experiments were performed on data collections that included single book corpus and multiple book corpus.• FindingsCombining keywords with certain semantic features achieves significant improvement of representation quality over the keywords-only approach in a relatively homogeneous single book corpus. Changing the ratios of the combined different features also affects the performance.While semantic mixtures can work well in single book corpus, they lose their increased effectiveness over keywords in the multiple-book corpus. This raises a concern about whether the semantic representations in the multiple book corpus are homogeneous and coherent enough to apply semantic features. The terminology issue among textbooks negatively impacts the ability of the SOM to generate a high quality map for heterogeneous collections.• Originality/valueWe explored the use of higher-level document representation features for the development of better-quality SOM. In addition, we piloted a specific method for evaluating the SOM quality based on the organization of information content in the map. IntroductionInformation maps (Kohonen, 1982) are becoming popular as interfaces to view and access large data collections such as digital libraries (DL). Unlike traditional search-based access, which provides selective and fragmented access to information, information maps allow users to comprehend large collections, to focus on the most interesting parts, and to explore specific resources in the context of their relationships to other resources and the library. Properties of information maps make them an excellent complement to search and browsing interfaces for DL. A recent study comparing student use of search, browsing and information map interfaces in an educational DL found that information maps were the method most preferred by students for accessing information; they were four times more popular than traditional search-based access methods. Several kinds of maps have been explored as interfaces to access large collections of resources (Börner and Chen, 2002, Yang et al., 2003, Dang et al., 2009, Perugini et al., 2004. Among these approaches, A self-organizing Map (SOM) (Kohonen, 1982) is frequently considered to be the most promising mapping approach for large document collections. While being most popular as a tool for two-dimensional cl...

show abstract

“…Lo and Bavarian addressed the effect of neighborhood function selection on the rate of convergence of the SOM algorithm [16]. Kiang et aldeveloped a "circular" training algorithm that tries to overcome some of the ineffective topological representations caused by the "boundary" effect [17]. Fritzke proposed a new self-organizing neural network model that can determine shape as well as size of the network during the simulation in an incremental fashion [18].…”

Section: Introductionmentioning

confidence: 99%

Fast self-organizing feature map algorithm

Chang

2000

IEEE Trans. Neural Netw.

View full text Add to dashboard Cite

We present an efficient approach to forming feature maps. The method involves three stages. In the first stage, we use the K-means algorithm to select N2 (i.e., the size of the feature map to be formed) cluster centers from a data set. Then a heuristic assignment strategy is employed to organize the N2 selected data points into an N x N neural array so as to form an initial feature map. If the initial map is not good enough, then it will be fine-tuned by the traditional Kohonen self-organizing feature map (SOM) algorithm under a fast cooling regime in the third stage. By our three-stage method, a topologically ordered feature map would be formed very quickly instead of requiring a huge amount of iterations to fine-tune the weights toward the density distribution of the data points, which usually happened in the conventional SOM algorithm. Three data sets are utilized to illustrate the proposed method.

show abstract

Improving the effectiveness of self-organizing map networks using a circular Kohonen layer

Cited by 9 publications

References 22 publications

Devanagari Handwritten Digits Recognition using Weighted Neighborhood Self-Organizing Map

Devanagari Handwritten Digits Recognition using Weighted Neighborhood Self-Organizing Map

Improving self‐organising information maps as navigational tools: a semantic approach

Fast self-organizing feature map algorithm

Contact Info

Product

Resources

About