Landmark map: An extension of the self-organizing map for a user-intended nonlinear projection

Onishi, Akinari

doi:10.1016/j.neucom.2019.12.125

Cited by 8 publications

(6 citation statements)

References 29 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The self-organizing map [14] is one of the most popular unsupervised learning algorithms for non-linear data projection for visualization. The self-organization map and its variants have many applications [31,32,33,34,35]. Inspired by V1 lateral influences we also design a modified version of the Self-organizing map algorithm.…”

Section: V1 Inspired Som Modelmentioning

confidence: 99%

An anatomy-based V1 model: Extraction of Low-level Features, Reduction of distortion and a V1-inspired SOM

Roy¹,

Pal²

2023

Preprint

View full text Add to dashboard Cite

We present a model of the primary visual cortex V1, guided by anatomical experiments. Unlike most machine learning systems our goal is not to maximize accuracy but to realize a system more aligned to biological systems. Our model consists of the V1 layers 4, 2/3, and 5, with inter-layer connections between them in accordance with the anatomy. We further include the orientation selectivity of the V1 neurons and lateral influences in each layer. Our V1 model, when applied to the BSDS500 ground truth images (indicating LGN contour detection before V1), can extract low-level features from the images and perform a significant amount of distortion reduction. As a follow-up to our V1 model, we propose a V1-inspired self-organizing map algorithm (V1-SOM), where the weight update of each neuron gets influenced by its neighbors. V1-SOM can tolerate noisy inputs as well as noise in the weight updates better than SOM and shows a similar level of performance when trained with high dimensional data such as the MNIST dataset. Finally, when we applied V1 processing to the MNIST dataset to extract low-level features and trained V1-SOM with the modified MNIST dataset, the quantization error was significantly reduced. Our results support the hypothesis that the ventral stream performs gradual untangling of input spaces.

show abstract

Section: V1 Inspired Som Modelmentioning

confidence: 99%

An anatomy-based V1 model: Extraction of Low-level Features, Reduction of distortion and a V1-inspired SOM

Roy¹,

Pal²

2023

Preprint

View full text Add to dashboard Cite

show abstract

“…The Kohonen network, namely Self-Organization Feature Map (SOFM), is a selforganizing competitive neural network proposed by Kohonen et al in 1981, which is an unsupervised learning model [26]. The Kohonen network is a neural network of an input layer and a competing layer (output layer) that realizes the bidirectional link between two layers through a full connection.…”

Section: Kohonen Networkmentioning

confidence: 99%

SSKM_DP: Differential Privacy Data Publishing Method via SFLA-Kohonen Network

Chu

et al. 2023

Applied Sciences

View full text Add to dashboard Cite

Data publishing techniques have led to breakthroughs in several areas. These tools provide a promising direction. However, when they are applied to private or sensitive data such as patient medical records, the published data may divulge critical patient information. In order to address this issue, we propose a differential private data publishing method (SSKM_DP) based on the SFLA-Kohonen network, which perturbs sensitive attributes based on the maximum information coefficient to achieve a trade-off between security and usability. Additionally, we introduced a single-population frog jump algorithm (SFLA) to optimize the network. Extensive experiments on benchmark datasets have demonstrated that SSKM_DP outperforms state-of-the-art methods for differentially private data publishing techniques significantly.

show abstract

“…However, although the generalized U-matrix is able to visualize the similarities and dissimilarities among high-dimensional data points in a scatter plot of the projected points, it is unable to visualize the disruption of clusters, based on which the quality of structure preservation is defined [25]. An extension of the SOM that utilizes several landmarks, e.g., pairs of nodes and data points (referred to as LAMA) is proposed in [26]. LAMA generates a user-intended nonlinear projection in order to achieve, e.g., the landmark-oriented data visualization.…”

Section: Related Recent Workmentioning

confidence: 99%

“…LAMA generates a user-intended nonlinear projection in order to achieve, e.g., the landmark-oriented data visualization. However, missetting of the LAMA's learning parameters, which are manually adjusted, may cause the mesh grid of codebook vectors to be twisted or wrinkled [26]. An emergent SOM (ESOM) concept is proposed in [25].…”

Section: Related Recent Workmentioning

confidence: 99%

Evolution of SOMs’ Structure and Learning Algorithm: From Visualization of High-Dimensional Data to Clustering of Complex Data

Gorzałczany

Rudziński

2020

Algorithms

View full text Add to dashboard Cite

In this paper, we briefly present several modifications and generalizations of the concept of self-organizing neural networks—usually referred to as self-organizing maps (SOMs)—to illustrate their advantages in applications that range from high-dimensional data visualization to complex data clustering. Starting from conventional SOMs, Growing SOMs (GSOMs), Growing Grid Networks (GGNs), Incremental Grid Growing (IGG) approach, Growing Neural Gas (GNG) method as well as our two original solutions, i.e., Generalized SOMs with 1-Dimensional Neighborhood (GeSOMs with 1DN also referred to as Dynamic SOMs (DSOMs)) and Generalized SOMs with Tree-Like Structures (GeSOMs with T-LSs) are discussed. They are characterized in terms of (i) the modification mechanisms used, (ii) the range of network modifications introduced, (iii) the structure regularity, and (iv) the data-visualization/data-clustering effectiveness. The performance of particular solutions is illustrated and compared by means of selected data sets. We also show that the proposed original solutions, i.e., GeSOMs with 1DN (DSOMs) and GeSOMS with T-LSs outperform alternative approaches in various complex clustering tasks by providing up to 20 % increase in the clustering accuracy. The contribution of this work is threefold. First, algorithm-oriented original computer-implementations of particular SOM’s generalizations are developed. Second, their detailed simulation results are presented and discussed. Third, the advantages of our earlier-mentioned original solutions are demonstrated.

show abstract

Landmark map: An extension of the self-organizing map for a user-intended nonlinear projection

Cited by 8 publications

References 29 publications

An anatomy-based V1 model: Extraction of Low-level Features, Reduction of distortion and a V1-inspired SOM

An anatomy-based V1 model: Extraction of Low-level Features, Reduction of distortion and a V1-inspired SOM

SSKM_DP: Differential Privacy Data Publishing Method via SFLA-Kohonen Network

Evolution of SOMs’ Structure and Learning Algorithm: From Visualization of High-Dimensional Data to Clustering of Complex Data

Contact Info

Product

Resources

About