Methods and principles of building "computer systems that automatically improve with experience."

learn more… | top users | synonyms

0
votes
0answers
11 views

Choice of loss function in correlation matrix prediction

There is a random vector $X=(X_1,\ldots,X_p)$, with $p$ large, $E[X]=0$ and $V[X_j]=1\ \forall \, j=1,\ldots,p$, but the correlations are different from zero. We cannot assume multivariate normality ...
7
votes
3answers
198 views

how to represent geography or zip code in machine learning model or recommender system?

I am building a model and I think that geographic location is likely to be very good at predicting my target variable. I have the zip code of each of my users. I am not entirely sure about the best ...
3
votes
0answers
28 views

What is the posterior probability of the data given the model used for model averaging with Bayesian Logistic Regression?

I am trying to learn about Bayesian Model Averaging using Bayesian Logistic Regression (Genkin, A., Lewis, D. D., & Madigan, D. (2007). Large-scale Bayesian logistic regression for text ...
0
votes
0answers
14 views

Facing Single-Class Training Set when Using Random Sampling

In a highly imbalanced binary classification (rare class < 10% of whole data), when I perform random sample selection (less than 15% of whole data to be selected for training) in a trial of 1000 ...
0
votes
0answers
24 views

Hidden Markov Models methods for selecting optimal number of states

Package RHmm (R) I have a vector which I fit into a hmm model in an attempt to select an optimal number of states for a hidden markov model. ...
1
vote
1answer
20 views

Quantifying applicability domain for predictive models?

I have a big dataset and I want to build a classification model (svm, rf, ann etc.). Then I split the original dataset into training set and test set. I build the model using training set. After it ...
0
votes
0answers
24 views

Is a Gaussian-Gaussian RBM just a linear model?

The 'conventional' configuration of RBMs are Binary-Binary and Gaussian-Binary (and sometimes Binary-Gaussian) units. Although it is possible for both the visible and hidden units to be gaussian, ...
0
votes
0answers
23 views

Best way to classify a set through a single feature

I need to classify a single dataset through a numeric value. I added below a simple dataset to explain what I need. Restriction: Category has two values: 1 or ...
0
votes
0answers
13 views

Statistical tests to check quality of data before learning model

I have collected some gene expression data and I want to learn a Bayesian network out of it. Before that, I want to do some statistical analysis to test the quality of my data. Now I want to know ...
1
vote
0answers
14 views

Learning to Rank: query-dependent vs. query-independent features

I've been doing some reading about learning to rank - specifically lambdaMART - and one thing I am confused about is the role of features. When training a model, should one only use query-dependent ...
0
votes
0answers
21 views

Information gain from a data set

I have a doubt regarding the use of information gain for data classification rather than conventional term frequency matrix.Lets assume I have 500 documents and I created a feature set of 12500*500. ...
1
vote
1answer
29 views

Computational Complexity of Prediction using SVM and NN?

I've seen answers discussing the complexity of training SVMs and neural nets, but how about for predicting new responses once a model has been trained? For context, I'm working on an app that should ...
0
votes
1answer
33 views

Measuring Statistical Significance of Binary Classification using Matthews Correlation Coefficient

Based on the following relationship between Matthew's Correlation Coefficient (MCC) and Chi Square: (MCC is the Pearson product-moment Correlation Coefficient) Is it possible to conclude that: By ...
5
votes
1answer
106 views

Why is optimizing a mixture of Gaussian directly computationally hard?

Consider the log likelihood of a mixture of guassians: $$l(S_n; \theta) = \sum^n_{t=1}logP(x^{(t)}|\theta) = \sum^n_{t=1}log\sum^k_{i=1}p_iP(x^{(t)}|\mu^{(i)}, \sigma^2_i)$$ I was wondering why it ...
1
vote
1answer
26 views

Hidden Markov Model: Predict observation sequence from state sequence

Given a transition matrix, starting probability, means and covariances Is it possible to predict the most likely observed sequence for a given state sequence using the above details? If yes, how? ...
0
votes
0answers
20 views

Feature selection for one class SVM

I have around 300 features, i need to choose features for one class svm. can some one tell me the ideal algorithm for this use case. I know about that for feature selection regularised random ...
0
votes
0answers
19 views

Objective function of an SVM [on hold]

I have used the svm function in the e1071 package of R software to model my data using variables selected by my feature selection method. I have obtained predictions from this model using the ...
2
votes
1answer
41 views

Difference between Bayes network, neural network, Petri Nets and decision tree

What is the difference between Neural network, Bayesian network, Decision tree and Petri Nets eventhough they are all graphical models and visually depict cause-effect relationship. Thank you
1
vote
0answers
19 views

HMM learning from video data?

I am having a problem understanding how to learn the parameters for the HMM from observed data. Let's say that my HMM model has one hidden variable for affect(emotion) with three values/states (anger, ...
0
votes
1answer
48 views

Linear Regression Real Life Example

I am learning Machine Learning(Linear Regression) from Prof. Andrew's lecture. While listening when to use normal equation vs gradient descent, he says when our features number is very high(like 10E6) ...
2
votes
1answer
18 views

Why is the decision function for probabilistic models a quotient (when we only consider two models)?

Take for example, that we want to find the probabilistic model for only two document types (doc can be + or -). I was trying to understand why the way that we classify a document model was with the ...
0
votes
1answer
22 views

decision boundary of support vector machine when data is not linearly separable

Screenshot from this video: This describes the decision boundary of support vector machine as a optimization problem with two constraints. But it seems to assume that the data points are linearly ...
3
votes
2answers
141 views

How to do multivariate machine learning? (predicting multiple dependent variables)

I am looking to predict groups of items that someone will purchase... i.e., I have multiple, colinear dependent variables. Rather than building 7 or so independent models to predict the probability ...
0
votes
1answer
20 views

Some Basic things we need to do when we are doing text classification

I am working on a project where I have to do multi-label text classification. I want to understand that whether my approach is correct or I am missing something. I am using R to do it. Clean ...
1
vote
1answer
39 views

Decision trees for advertising data

Assuming a dataset with the following attributes: Date (truncated), f1 ... fn, ...
1
vote
0answers
18 views

How to derive errors in neural network with the backpropagation algorithm?

From this video by Andrew Ng around 5:00 How are $\delta_3$ and $\delta_2$ derived? In fact, what does $\delta_3$ even mean? $\delta_4$ is got by comparing to y, no such comparison is possible ...
0
votes
0answers
25 views

Performance comparsion study on real data in MATLAB (machine learning)

I need to compare performance using 2-class classifiers–an LDA classifier and Exact Bayes in MATLAB. I have to use this dataset. Can anyone give me any advice how to do that (at least the steps of the ...
-1
votes
1answer
21 views

Is it OK to increase validation checks and decrease min gradient while training neural network?

My input vector is a 130*85 matrix and my target vector is 130*26 matrix. I am using the below parameters for training the network with 60 hidden nodes. ...
1
vote
1answer
46 views

Can a machine learning algorithm be evaluated based on a random sample?

I am trying to evaluate how well (or bad) a semi-supervised algorithm is performing on a given dataset. The algorithms assigns one of 10 labels to each data point. The dataset is huge, and it's not ...
1
vote
2answers
45 views

Dealing with different time series data in Machine Learning

I am trying to create a stock market model based on fundamental variables for the US economy. I am using R. Some of the variables I am looking to include are: GDP, Unemployment Rate, Initial Claims, ...
0
votes
1answer
34 views

Classification Accuracy

I am classifying text based on news headlines and I am achieving accuracy up to approx 80%. I want to improve it more. But issue is that when I calculate the same with synonyms using the code below: ...
0
votes
0answers
19 views

Tutorial on Radial Basis Function Networks?

I want to learn about Radial Basis Function Neural Networks, can you please suggest a good introduction or tutorial? All the introductions I found are rather short or incomplete or so.
0
votes
0answers
24 views

Feature Selection for look alike modeling using k-NN

I have a list of items and various parameters for each items. For each item on my list i need to identify items which are similar to the item from my whole population . I am planning on using K-NN ...
0
votes
1answer
25 views

Decision boundary equation of the perceptron

As I know the standard linear equation has the following form in $R^2$: $w_1 x_1 + w_2 x_2 = b$ where $b$ is the intercept with $x_2$ Also I know that a decision boundary in $R^2$ for a perceptron ...
2
votes
1answer
32 views

algorithms to pick best collection of classifiers

I have a bunch of methods that classify a binary outcome. I'm trying to figure out if some combination of those classifiers is better than any others. I'm hoping to run a bunch of methods. I've ...
0
votes
0answers
7 views

How to validate classifier (built by using MLN method)?

I have developed a method (let's call it Method X) that has a classifier function. The classifier function was built by using MLN (Markov logic network). I need to ...
2
votes
3answers
67 views

Does using a kernel function make the data linearly separable?

I'm reading about SVM and I learned that we use a kernel function so the data become linearly separable in the high dimensional (vector?) space. But then I also learned that they use the soft-margin ...
-2
votes
0answers
42 views

Technologies behind Siri and Google Now?

I'm interested to know how Siri and Google Now works. What are the technologies behind them? I don't mean voice recognition (which is obvious) but the other stuff. Like interpreting the input and ...
0
votes
1answer
39 views

How to make stochastic gradient descent algorithm converge to the optimum?

(Background info taken from my blog) In logistic regression, the hypothesis function, which models the relationshiop between the dependent variable $P(y = 1)$ and the independent variable $X$, is : ...
2
votes
1answer
32 views

How can I evaluate the performance of a system that generates word clusters?

The word2vec tool uses deep learning to compute vector representations of words. They've mentioned that - "The word vectors can be also used for deriving word classes from huge data sets. This is ...
0
votes
1answer
35 views

Performing hierarchical clustering on a large data set

I have been applying complete linkage on about 5,000 points using matlab with no problem. I want to extend this method to much more elements. It would take me a long time to process my data to test ...
1
vote
0answers
20 views

Are these descriptions of batch gradient descent algorithm conflicting each other?

The first one is from Andrew Ng The second one is from Francis Bach I might be a little confused, but why is there a summation of partial derivatives in the second description and none in the ...
2
votes
1answer
41 views

Gradient decay in neural networks

I read that in traditional feed-forward neural nets the gradients in the early layers decay very quickly and that this is 'a bad thing'. But I don't understand why. Can someone please explain what ...
0
votes
0answers
24 views

What is multi run lasso regression?

I have problem in understanding of multi-run lasso regression. Basically, I know what is lasso regression, but don't know what is multi-run lasso regression, which sometimes I see literatures. Does ...
0
votes
2answers
26 views

Latent Dirichlet Allocation as input for WEKA

I am using the Weka API for my research about document classification. I wish to apply Latent Dirichelet Allocation on my dataset followed by using a classifier in Weka. However, it is not so clear to ...
6
votes
2answers
566 views

Meaning of a neural network as a black-box?

I often hear people talking about neural networks as something as a black-box that you don't understand what it does or what they mean. I actually I can't understand what they mean by that! If you ...
1
vote
0answers
32 views

Rough estimates for training time of deep belief networks

I'm still learning about deep learning. However I'm currently interested to know if deep learning architectures scale well or not. Suppose I have a dataset with 1 million training examples, can you ...
0
votes
0answers
10 views

Why feature maps are indexed by two indices?

I'm reading about convolutional neural networks. As I understood a feature map is a set of neurons (i.e like a single hidden layer in traditional ANN). So why feature maps are indexed by (i,j)? ...
0
votes
1answer
19 views

Updating set of probabilities for sampling with features importance

I'm currently working on some algorithm and I'm kinda out of idea for a problem I'm trying to tacle. Basically I'm trying to subsample the features of a dataset. I want to subsample that given this ...
1
vote
0answers
23 views

Compare averaged GLM with boosted regression trees using cross validation : d2 and RMSE calculation

I want to compare BRT and averaged glm models on test sets by calculating the explained deviance and RMSE. How can I calculate d2 and RMSE from predictions? I use the following functions: gbm1 ...