self-organizing-maps

问题 I have a large dataset I am trying to do cluster analysis on using SOM. The dataset is HUGE (~ billions of records) and I am not sure what should be the number of neurons and the SOM grid size to start with. Any pointers to some material that talks about estimating the number of neurons and grid size would be greatly appreciated. Thanks! 回答1: I don't have a reference for it, but I would suggest starting off by using approximately 10 SOM neurons per expected class in your dataset. For example,

问题 I have a large dataset I am trying to do cluster analysis on using SOM. The dataset is HUGE (~ billions of records) and I am not sure what should be the number of neurons and the SOM grid size to start with. Any pointers to some material that talks about estimating the number of neurons and grid size would be greatly appreciated. Thanks! 回答1: I don't have a reference for it, but I would suggest starting off by using approximately 10 SOM neurons per expected class in your dataset. For example,

问题 i've been trying to modify the program so that it could accept more than one data for a single alphabet character for example letter "A". there were some sort of ContainsKey function that allow only one key from keyboard to hold only one data. how to make it possible to hold more than one data? I'm gonna make it very clear, this is an online OCR program using unsupervised neural network. when a user draw a character in the drawing space, they will have the option to add the character into the

问题 I am playing a bit with the SOMbrero package. I would like to attach the cluster numbers created like so (taken from here): my.sc <- superClass(iris.som, k=3) and X and Y coordinates of the SOM nodes to the training dataset. In some code, where I use the kohonen package, I create clusters like this: range01 <- function(x){(x-min(x))/(max(x)-min(x))} ind <- sapply(SubsetData, is.numeric) SubsetData[ind] <- lapply(SubsetData[ind], range01) TrainingMatrix <- as.matrix(SubsetData) GridDefinition

问题 I am playing a bit with the SOMbrero package. I would like to attach the cluster numbers created like so (taken from here): my.sc <- superClass(iris.som, k=3) and X and Y coordinates of the SOM nodes to the training dataset. In some code, where I use the kohonen package, I create clusters like this: range01 <- function(x){(x-min(x))/(max(x)-min(x))} ind <- sapply(SubsetData, is.numeric) SubsetData[ind] <- lapply(SubsetData[ind], range01) TrainingMatrix <- as.matrix(SubsetData) GridDefinition

问题 I have implemented the Self-Organizing Map(SOM) algorithm in MATLAB. Suppose each of the data points are represented in 2-dimensional space. The problem is that I want to visualize the movement of each of the data points in the training phase i.e. I want to see how the points are moving and eventually forming clusters as the algorithm is in progress say at every fix duration. I believe that this can be done through Simulation in MATLAB,but I don't know how to incorporate my MATLAB code for

问题 I have a text file that include data. My text file: young, myopic, no, reduced, no young, myopic, no, normal, soft young, myopic, yes, reduced, no young, myopic, yes, normal, hard young, hyperopia, no, reduced, no young, hyperopia, no, normal, soft young, hyperopia, yes, reduced, no young, hyperopia, yes, normal, hard I read my text file load method %young=1 %myopic=2 %no=3 etc. load iris.txt net = newsom(1,[1 5]); [net,tr] = train(net,1); plotsomplanes(net); Error code: ??? Undefined

问题 I've looked at a lot of theoretical examples of SOMs, but one thing is not really clear to me: is the location of nodes dependent on their weights? For example, will nodes with a larger weight be on one side of the map, while nodes with a smaller weight will be further away on the map? 回答1: No. In an SOM (aka Kohonen Map) the weight function is applied to your data not the the "Neurons". Weights are used during map construction (training), i.e., calculated at each iteration and for each

问题 How exactly is an U-matrix constructed in order to visualise a self-organizing-map? More specifically, suppose that I have an output grid of 3x3 nodes (that have already been trained), how do I construct a U-matrix from this? You can e.g. assume that the neurons (and inputs) have dimension 4. I have found several resources on the web, but they are not clear or they are contradictory. For example, the original paper is full of typos. 回答1: A U-matrix is a visual representation of the distances

问题 Could someone please provide some information on how to properly combine a self organizing map with a multilayer perceptron? I recently read some articles about this technique in comparison to regular MLPs and it performed way better in prediction tasks. So, I want to use the SOM as front-end for dimension reduction by clustering the input data and pass the results to an MLP back-end. My current idea of implementing it is it to train the SOM with a couple of training sets and to determine the