Multiple Back-Propagation Software - Version 2.2 available

Version 2.2 of Multiple Back-Propagation (MBP) is available for download either at http://dit.ipg.pt/MBP or alternatively at http://sourceforge.net/projects/mbp/ This version supports training networks with data containing missing values (data must be on a CSV file). A value is considered as being absent (missing) if its value is not present or if it is equal to a question mark (?). To deal with missing values MBP does not use imputation or any other pre-processing technique, but it rather creates a network with =93selective inputs=94, which can handle directly missing values. Selec

CFP: International Symposium on Open Source Intelligence & Web Mining 2010

International Symposium on Open Source Intelligence & Web Mining 2010 http://www.osint-wm.org/ After the great success of OSINT-WM 2008 and OSINT-WM 2009, which were held in conjunction with the 12th International Conference Information Visualization, IV08 in London, UK and 13th International Conference Information Visualization, IV09 in Barcelona, Spain, we look forward to a more successful OSINT-WM 2010 http://osint-wm.org/index.html to be held in Conjunction with the ASONAM 2010. The aim of this symposium is to provide an international forum for researchers, profession

Best possible classification rate on the PROBEN1 / HEART1 data set

I am trying out some machine learning techniques on the PROBEN1 benchmark, specifically the HEART1 test set. I noticed that several papers report results with about 18-20% test error. Is it possible to achieve better classification rates? Kind Regards Theo

Some basics

I've studied NN's for many years and even played with a few. However, I cannot make them work in a real experiment because I just can't get my head around translating real-world values into some kind of across-the-board standard values that the NN's use. For example, how would one convert a picture of just a face into a value "between 0 and 1" or "-1 and 1," while keeping relative consistency with translating engine rpm to "0 and 1," and wind speed "between 0 and 1?" Hope this makes sense. Sometimes I wonder if it matters if there IS consistency across the data domains wh

EasyNN-plus V13 free update

Coming in the next update. Changes since the half price offer. The update will be free to all users of version 13 including those who bought it in the half price offer. 1. Fixed the crash when right click on empty grid. 2. Disabled Validating Rules when grid is empty. 3. Fixed a looping problem in View > Graph > Column Values after Undo. 4. Fixed the problems with setting and executing a function on a grid range. 5. Added more detail to messages when importing xls files. 6. Added warning message for missing headers when importing xls files. 7. Removed right click Se

Call for Papers: 2010 International Conference on Computer Applications and Industrial Electronics (ICCAIE 2010)

We apologize if you receive multiple copies of this message. Appreciate it if you can disseminate to your peers. Dear Sir/Madam, ======================= CALL FOR PAPER ======================= It is our great pleasure to announce the 2010 International Conference on Computer Applications and Industrial Electronics (ICCAIE 2010) to be held in Kuala Lumpur, MALAYSIA. between the 5th-8th December 2010. The conference will be organized to invite international delegates, to share their latest research findings besides getting to know Malaysia. ICCAIE 2010 will be organized by the

Bootstrapping an ANN

Dear neural netters, I have a question regarding the bootstrapping of an ANN. The ANN applied in the context of my study used a back-propagation algorithm (nevprop) that consisted of five input variables, four hidden layers containing five neurons each, and one output layer. My inputs were standardized (0 to 1) to facilitate inclusion of an age variable and weights were initially set using a random seed value. Optimization of the network was undertaken using bootstrapping methods, with 50% of the training data removed from the training set and used to assess the accuracy of the ne

1st announcement: WSOM 2011, Workshop on Self-Organizing Maps

8th WORKSHOP ON SELF-ORGANIZING MAPS WSOM 2011, Espoo, Finland Venue: Aalto University School of Science and Technology and Dipoli Conference Center 13-15 June 2011 GENERAL INFORMATION WSOM 2011 will bring together researchers and practitioners in the field of self-organizing systems, with a particular emphasis on the self-organizing maps. It will highlight key advances in these and closely related fields. WSOM 2011 is the eighth conference in a series of bi-annual international conferences started with WSOM'97 in Helsinki. The event will be co-located with the ICANN 2011 c

five year old obituary notice

On Thu, 25 Feb 2010, News Source <anonymous@not-for-mail.invalid> wrote: >Google Groups: How Not to Run an Archive > >http://www.kevinalfredstrom.com/2010/02/google-groups-how-not-to-run-an-archive/ > >by Kevin Alfred Strom > >IT'S BEING acknowledged everywhere - now by Wired magazine and the >SEO Theory newsletter. Google Groups is totally broken. > >http://www.wired.co.uk/news/archive/2009-10/08/the-lost-archives-of-google-groups.aspx > >http://www.seo-theory.com/2008/12/30/how-google-groups-search-has-become-virtually-worthless/ > >It began well. Google acquired the Deja

2012: It's Only The End Of The World

-----BEGIN PGP SIGNED MESSAGE----- As the pseudochristians and other Antichrist Atheists around the Fallen world scramble to avoid the imminent universal desolation, it's gratifying to watch this international Empire of Antichrist being universally destroyed more of flesh and bone than of walls and stone. And there's absolutely nothing they can do to stop it. Their universal desolation is inevitable unto the culmination of

There is a problem in MATLAB 2009a using divideint function.

Hello people. There is a problem when we use the divide param divideint to train with early stopping. Validation ratio is used in test set and test ratio is used in validation set. You can correct it modifying two lines in divideint.m file. Move line 107 to line 106 and vice-versa. Till next time! Rub=E9n.

Call for Papers: The 2010 International Conference of Manufacturing Engineering and Engineering Management (ICMEEM 2010)

CFP: The 2010 International Conference of Manufacturing Engineering and Engineering Management (ICMEEM 2010) From: IAENG - International Association of Engineers http://www.iaeng.org/WCE2010/ICMEEM2010.html Draft Paper Submission Deadline: 6 March, 2010 Camera-Ready papers & Registration Deadline: 31 March, 2010 WCE 2010: London, U.K., 30 June -2 July, 2010 The conference ICMEEM=9210 is held under the World Congress on Engineering 2010. The WCE 2010 is organized by the International Association of Engineers (IAENG), a non-profit international association for the engineers and the

The Greater Good

-----BEGIN PGP SIGNED MESSAGE----- Have you noticed how the evildoers of the world are working toward the greater good, wholly regardless of their motives? This is because the power of God is unlimited; whereas, the power of the Fallen, sinful man is limited by the aforesaid. ___________________________________________________________ Friday December 21st 2012 AD is 1039 days away and counting. The final worldly pope BXVI is 30,256 days old and counting.

CMOS Circuit Design openings with a leading firm in Bangalore, India

Hi, We have an excellent opening for our below requirement =96 Circuit Design with one of our leading and reputed clients into product development and research and development activities in Bangalore, India. Please find below the details: Custom Circuit Design -: Active hands on experience in CMOS Custom VLSI Circuit Design using industry leading tools (CADENCE preferred) Thorough understanding of CMOS technology and devices. Familiar with VHDL, schematics entry, circuit simulation (SPICE), custom layout, layout verification methodologies, design for yield and manufacturability,

Data training with backpropagation

Hi, i would like to train network using backpropagation in matlab with large data where the network structure is 5 nodes for input 6 nodes hideen 4 nodes output. i should recognize data of four class ( A ,B,C,D ) some of our data are as follows: (A) (B) (C) (D) 1 1 1 1 1 1 1 1 -0.00281 -0.00276 -0.00248 0.019727 -0.000696 -0.002768 -0.001826 -0.001998 -0.09323 -0.09280 -0.09280 -0.090197 -0.093245 -0.093217 -0.093170 -0.0

Call for Papers: The 2010 International Conference of Financial Engineering (ICFE 2010)

CFP: The 2010 International Conference of Financial Engineering (ICFE 2010) From: IAENG - International Association of Engineers http://www.iaeng.org/WCE2010/ICFE2010.html Draft Paper Submission Deadline: 6 March, 2010 Camera-Ready papers & Registration Deadline: 31 March, 2010 WCE 2010: London, U.K., 30 June -2 July, 2010 The conference ICFE=9210 is held under the World Congress on Engineering 2010. The WCE 2010 is organized by the International Association of Engineers (IAENG), a non-profit international association for the engineers and the computer scientists. Our congress co

EasyNN-plus 13.0a

EasyNN-plus V13.0a is available from today. The main change is as follows:- Functions can be added to Nodes and Layers of nodes. They are executed when the nodes are processed while the network is learning the training information in the grid. Some functions can decrease the number of cycles needed for the network to converge. For example if a hidden layer function is added to multiply the activation it can improve learning speed by 2 or 3 times. -- Neural Planner Software Ltd www.NPSL1.com EasyNN-plus. Neural Networks plus. www.easynn.com SwingNN.

Scaled Conjugate Gradient in Netlab

Hi, I am using scg function of netlab to train a mlp network. I want to set the lambda and sigma parameter before training starts. The function is scg(f, x, options, gradf, varargin). Options is a vector of parameters. I am not sure which parameter is lambda and which is sigma. The help manual does not give information about this. How should I set the parameters for lambda and sigma? Many thanks David

pattern classification with mlp

hi everybody; i have a problem in training in mlp, i read posted about this but i dont know improve performance of may question, i have 6 feature and 10 class and i have 32 training sample , i use this code for training data but i have not a good result, just 11 sample classified corrected, input=r'; %32*6 output=T'; %32*10 net = newff(minmax(input),[10 10],{'losig' 'logsig'},'trainlm'); net.trainParam.epochs = 100; net.trainParam.goal = 1e-10; net = train(net,input,output); it train one epoch "TRAINLM, Epoch 1/100, MSE NaN/1e-010, Gradient NaN/1e-010 TRAINLM, Max

Call for Papers: The 2010 International Conference of Applied and Engineering Mathematics (ICAEM 2010)

CFP: The 2010 International Conference of Applied and Engineering Mathematics (ICAEM 2010) From: IAENG - International Association of Engineers http://www.iaeng.org/WCE2010/ICAEM2010.html Draft Paper Submission Deadline: 6 March, 2010 Camera-Ready papers & Registration Deadline: 31 March, 2010 WCE 2010: London, U.K., 30 June -2 July, 2010 The conference ICAEM=9210 is held under the World Congress on Engineering 2010. The WCE 2010 is organized by the International Association of Engineers (IAENG), a non-profit international association for the engineers and the computer scientists

High Level Encryption to Protect Your Patient and Research Data

High Level Encryption to Protect Your Patient and Research Data The Kyrptonite series of programs uses a high-level encryption algorithm. The algorithm and method was tested by the National Security Agency, and was one of the few programs originating outside of the agency considered for additional testing and evaluation. Due to the advanced algorithm of the program, this program is limited to sales inside the United States only due to evaluation by the Bureau of Industry and Security of the department of commerce. The Kryptonite-3 program allows you to have 3 keys that can be arra

SCface - Surveillance Cameras Face Database

Dear colleagues, We would like to inform about a new face database: SCface - Surveillance Cameras Face Database. The database is available free of charge to researchers, and obtainable through the procedure described at the SCface web-site: http://www.scface.org/ Here is a short description of the database, as presented on the above web-site: SCface is a database of static images of human faces. Images were taken in uncontrolled indoor environment using five video surveillance cameras of various qualities. Database contains 4160 static images (in visible and infrared spectru

The internal workings of the human neuron

What is the current scientific status of the idea that a human neuron could be a quantum computer? I would recall that this idea arose in the 1990's, but I have not been following the scientific discussion concerning that thought. kind regards, Mr Antti J. Ylikoski Helsinki, Finland, the E.U. (I posted this to the comp.ai newsgroup, which is moderated... The moderators rejected it. Have I become a persona non grata here?) Idem

Multiple Back-Propagation Version 2.1.2 is now available

Multiple Back-Propagation is a free software application (released under GPL v3 license) for training neural networks with the Back- Propagation and the Multiple Back-Propagation algorithms. The latest version of Multiple Back-Propagation can be downloaded at http://dit.ipg.pt/MBP/Download.aspx Version 2.1.2 is now available for download. This version supports CSV (Comma-separated values) files. CSV files can easily be generated by many programs such as Excel, thus CSV files are now the preferred method of input data (the other formats remain only for compatibility purposes) to Mult

SOM, automatic cluster boundaries detection?

Hi, I've been searching a lot for methods which automatically detect clusters boundaries ( if any ) with SOM. The vast majority of references resemble on the analyst expertise to visually take the decision in the number of the clusters and where are their boundaries. One exception so far was this reference Determining Cluster Boundaries using Particle Swarm Optimization Anurag Sharma, and Christian W. Omlin Anyone has other references and comments about that?