Building a Deep Learning Model for Process Optimisation

The objective of this paper is to present the process of building a Deep Learning Model for optimising the output for a Production Process from a Training sample using Weka Multilayer Perceptron. The scope is limited to implementation only and does not cover the theory behind Artificial Neural Networks. The Genetic Algorithm is specifically developed and adapted for the kind of data involved and is dealt with in detail This work is the outcome of a comprehensive prototyping and proof-of-concept exercise conducted at Turing Point (http://www.turing-point.com ) a consulting company focused on providing genuine Enterprise Machine Learning solutions based on highly advanced techniques such as 3D discrete event simulation, deep learning and genetic algorithms.


Regression model with auto correlated errors – Part 2, the models

In the first part I created the data for testing the Astronomical/Astrological Hypotheses. In this part, I started by fitting a simple linear regression model.


Break the Markov Chains of Oppression: Modeling without MCMC

You’ve seen the articles that say ‘MCMC is easy! Read this!’ and by the end of the article you’re still left scratching your head. Maybe after reading that article you get what MCMC is doing… but you’re still left scratching your head. ‘Why?’


Deep Learning Can be Applied to NLP

There is an article going around the rounds at LinkedIn that attempts to make an argument against the use of Deep Learning in the domain of NLP. The article written by Riza Berkan “Is Google Hyping it? Why Deep Learning cannot be Applied to Natural Languages Easily” has several arguments about DL cannot possibly work and that Google is exaggerating its claims. The latter argument is of course borderline conspiracy theory. Yannick Vesley has written a rebuttal “Neural Networks are Quite Neat: a Reply to Riza Berkan” where he makes his arguments on each point that Berkan makes. Vesley’s points are on the mark, however one can not ignore the feeling that DL theory has a few unexplained parts in it. However, before I do get into that, I think it is very important for readers to understand that DL currently is an experimental science. That is, DL capabilities are actually discovered by researchers by surprise. There are certainly a lot of engineering that goes into the optimization and improvement of these machines. However, its capabilities are ‘unreasonably effective’, in short, we don’t have very good theories to explain its capabilities.


on Cheap Learning: Partition Functions and RBMs

This is the question posed by a recent article. Deep Learning seems to require knowing the Partition Function-at least in old fashioned Restricted Boltzmann Machines (RBMs). Here, I will discuss some aspects of this paper, in the context of RBMs.


Why does Deep Learning work?

This is the big question on everyone’s mind these days.


Gelman’s MrP in R – What is this all about?

Multilevel regression with poststratification (MrP) is a useful technique to predict a parameter of interest within small domains through modeling the mean of the variable of interest conditional on poststratification counts. This method (or methods) was first proposed by Gelman and Little (1997) and is widely used in political science where the voting intention is modeling conditional on the interaction of classification variables. The aim fo this methodology is to provide reliable estimates on strata based on census counts. For those who have some background on survey sampling, this method should look very similar to the Raking method, where sampling weights are adjusted due to known census cell counts. However, a significant difference with Raking is that MrP is a model-based approach, rather than a design-based method. This way, even in the presence of a (maybe complex) survey design, MrP does not take it into account for inference. In other words, sampling design will be considered as ignorable. So, the probability measure that governs the whole inference is based on modeling the voting intention (variable of interest) to demographic categories (auxiliary variables).


Multiple Regression (Part 1)

In the exercises below we cover some material on multiple regression in R. If you obtained a different (correct) answer than those listed on the solutions page, please feel free to post your answer as a comment on that page. We will be using the dataset state.x77, which is part of the state datasets available in R. (Additional information about the dataset can be obtained by running help(state.x77).)


Feature Selection in Machine Learning (Breast Cancer Datasets)

Machine learning uses so called features (i.e. variables or attributes) to generate predictive models. Using a suitable combination of features is essential for obtaining high precision and accuracy. Because too many (unspecific) features pose the problem of overfitting the model, we generally want to restrict the features in our models to those, that are most relevant for the response variable we want to predict. Using as few features as possible will also reduce the complexity of our models, which means it needs less time and computer power to run and is easier to understand.


Structural Equation Modelling in R (Part 2)

This is the second part in a series on three articles about Structural Equation Modelling (SEM). This time I am glad to announce Jodie Burchell as a co-writer! In Structural Equation Modelling in R (Part 1) I explained the basics of CFA. SEM was explained as a general case of CFA that was going be explained later, so here we go.


Animated barplot and google map with R

It might happen that you will need a animated graph of any kind. For purposes of plotting demographic data and changes through the years, Google Maps and plotting maps, merging and converting jpg files into a animated gif, sure will give a nice visualization effect. Here is a sample of changes over the time period of three years on some dataset of my home town and graph can tell little bit more as numbers ??