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

learn more… | top users | synonyms (1)

0
votes
1answer
31 views

Standardization before applying ANOVA?

I have a matrix where the rows are the data points (samples) and the columns are the features. It is a multiclass (4 classes) problem. On this data I want to apply machine learning classifiers. But ...
1
vote
1answer
25 views

The relationship between adaboost and gradient boosting

I am reading the chapter 10 of "The Elements of Statistical Learning 2nd ed, (ESLII)", where the Adaboost algorithm is explained by minimizing the exponential loss using stagewise additive modelling ...
52
votes
2answers
3k views

Help me understand Support Vector Machines

I understand the basics of what a Support Vector Machines' aim is in terms of classifying an input set into several different classes, but what I don't understand is some of the nitty-gritty details. ...
0
votes
1answer
15 views

Why we need to extract a lot of features from a dataset for classification

I am newbie in machine learning. I have been studying about features extraction and some classification approaches, in the term of my study, I have a question in my mind, what the reasons we need to ...
0
votes
0answers
10 views

Running repeated cross-validation for multiple models using same dataset (caret package)

I'm currently using the train() function in the caret package to run 10-fold repeated cv on a random forest model. I would also like to explore other statistical and machine learning models for use ...
3
votes
1answer
786 views

How do I calculate random baseline?

I am a bit confused as to how to calculate random baseline. If I understand correctly the random baseline is calculated by adding up the squared probabilities of all the classes. The random baseline ...
0
votes
1answer
81 views

mixing binary and real-valued features with SGD

I'm going to be using a logistic regression model and using SGD to determine the feature weights. Is it OK for me to use a mix of binary and real features, without doing anything like scaling or ...
0
votes
0answers
11 views

How to extend independent significance features test to multiple classes?

I have found this post about feature selection using the independent significance features test. There is also an implementation provided. Unfortunately it only works for 2 classes but I have 4 ...
1
vote
1answer
686 views

Multidimensional scaling using Python

I have 6,000 points for which I have all pairwise distances in a distance matrix. I want to get an idea whether these data were generated by a mixture of Gaussian distributions so I'm trying to get a ...
1
vote
1answer
43 views

Training a random forest in R with a maximum false positive rate

I ran the following code in R: rf.classifier.master <- randomForest(my_response ~ ., data=feature.matrix) print(rf.classifier.master) and got the following ...
1
vote
0answers
19 views

Where to find Wikipedia's wiki.dat for LDA, an example used in Vowpal Wabbit? [on hold]

I am following Vowpal Wabbit's tutorial on LDA (Latent Dirichlet Allocation) found here. wiki.dat is used in it as a Wikipedia corpus or supplementary Wikipedia dictionaries, maybe. It must be ...
1
vote
1answer
18 views

Pre-training in deep convolutional neural network?

Have anyone seen any literature on pre-training in deep convolutional neural network? I have only seen unsupervised pre-training in autoencoder or restrcited boltzman machines.
1
vote
0answers
16 views

Perceptron Learning Algorithm: what is the probability that the viewed data is linearly separable, after some number of steps?

My understanding is that the PCA: will not converge if the data is not linearly separable might take exponentially many iterations, even if the data is linearly separable I'm wondering if, after ...
7
votes
3answers
539 views

How are classifications merged in an ensemble classifier?

How does an ensemble classifier merge the predictions of its constituent classifiers? I'm having difficulty finding a clear description. In some code examples I've found, the ensemble just averages ...
0
votes
0answers
19 views

Label outliers for anomaly detection

I am trying to detect anomalies using unsupervised learning techniques. However, I have the problem that it is impossible to generate controlled anomalies to use as a test set. My idea is to discover ...
0
votes
1answer
15 views

Do I need data standardization before t-SNE?

In my knowledge, t-SNE use sphere or diagonal normal distribution to represent the distance between point. So I guess that the distance is biased by the scale of respective features. To my surprise, ...
1
vote
0answers
18 views

Why multivariate distribution is used to model weights (polynomial coefficients) for prior distribution?

Lately, I have been reading Chris Bishop's work on Bayesian linear regression. What was difficult to understand is why is he models $w_0$ and $w_1$ using multivariate Gaussian distribution in prior ...
4
votes
1answer
67 views
+50

Relative variable importance for Boosting

I'm looking for an explanation of how relative variable importance is computed in Gradient Boosted Trees that is not overly general/simplistic like: The measures are based on the number of times a ...
1
vote
1answer
11 views

What should be the covariance matrices and weights for initializing EM/GMM with kmeans?

It's typical to initialize EM for Gaussian Mixture Models using the result of kmeans clustering. However, kmeans only gives you the means (centers) of the starting GMM, but EM initialization often ...
0
votes
0answers
9 views

How can I obtain the Simulations in CausalImpact package? [on hold]

Currently we are using your package CausalImpact to evaluate the effect of different interventions over the accident occurrence in different firms. I address to you in order to ask you if there is any ...
0
votes
1answer
60 views

PCA vs. Linear Regression

Am taking Ng's Machine Learning class on Coursera and in the below slide he distinguishes PCA from Linear Regression. He says that in Linear Regression, we draw vertical lines from the data points ...
1
vote
2answers
41 views

“Non-naive” bayesian classification algorithms

Based on the problem description in this post: Relating parameters to a measured variable Based on a suggestion, I thought of studying the relationship between the parameters and a measured metric ...
0
votes
0answers
13 views

How to perform Sensitivity Analysis of Bayesian network using R?

I know there are similar techniques for Sensitivity Analysis for Random Forests. I am looking for something similar for Bayesian Networks. I have built the Bayesian Network using bnlearn package in R. ...
-1
votes
0answers
15 views

What is the difference between a conditional random field model and a particle filter?

Please can any one explain the difference between a CRF model and a particle filter?
1
vote
1answer
412 views

10 fold cross validation model in weka

Trying to build a specific Neural Network arcitecture and testing it using 10 fold cross validation of a dataset. Now building the model is a tedious job and Weka expects me to make it 10 times for ...
0
votes
0answers
13 views

Issues in Testing SMO SVM

I am new be in SVM and SMO algorithm, I implemented SMO using the pseudocode provided in : “Fast training of support vector machines using SMO” by John platt. I am finding issues testing my ...
0
votes
1answer
62 views

Which language best to use for Machine Learning library? [on hold]

We have a body of theoretical work on nearest neighbors that we would like to implement and make the code publicly available. Question: what's the best language to use? We're considering java, python, ...
1
vote
2answers
27 views

Online learning that “forgets” older aspects learned? (short-term memory)

I am looking for an online learning classifier that is highly adaptable and has only short-term memory. I need such a think in a object tracking system with high-dimensional feature vectors. Maybe a ...
0
votes
0answers
20 views

Classifier that learns provided by only positive examples?

I was wondering if any of you has ever worked with classification/regression using only positive examples (one class). I would need such a system. The basic idea is that it is going to accurately ...
0
votes
1answer
205 views

Machine learning with trinomial features

I have 100,000 students who have each answered some multiple choice questions. Given their performance I want to work out what the chances are of a particular student answering the next question ...
1
vote
0answers
29 views

Tennis Analytics: How to Build Model Predicting Player Service Point Win %

I have collected a large amount of tennis match data including player names, court surface, player ranking points at time of match, handedness of player, point by point breakdown of match etc. I ...
0
votes
0answers
16 views

How large should the sample be to start using batch gradient descent versus normal equation

Suppose you want to train a multivariate linear regression on an n x m dataset. The runtime of determining your parameter theta = (theta0, theta1, ..., thetam) using the normal equations is ~ O(m^3), ...
0
votes
0answers
17 views

How do I “split” Gaussian mixture components when training EM/GMM based classifier?

In order to improve performance of my Gaussian Mixture Model based classifier, I was recommended to start with a single multivariate Gaussian, estimate its parameters, and "split" it into two ...
3
votes
0answers
25 views

SVM with non-negative weights

An SVM classifier can be obtained by solving the following, $\arg\min \frac{1}{2}\|W\|_2^2 + C\sum_i \max(0, 1-y_i (W^T\mathbf{x}_i + b))$ where $W$ is the hyperplane (or weights), $b$ is the bias, ...
1
vote
1answer
21 views

does sklearn rbm scale well with sparse high dimensional features

i am using scikit learn's RBM implementation. There are two problems: The running time is O(d^2) where d is the number of features. This becomes a problem in using high dimensionality sparse ...
0
votes
0answers
13 views

How to add cluster centers to the already transformed arrays with T-SNE Scikit Learn?

let's get this scikit original code, which is basically the one I'm using. My X is 2000x100 and in order to plot the clusters (plot on the right) I want to transform it with with the TSNE algorithm ...
0
votes
0answers
7 views

normalized mutual information implantation in java for community detection in graph range is not between 0 and 1 [on hold]

I write a program for calculating normalized mutual information for evaluate community detection. but i get values above 1 for nmi. normally it should be between 0 and 1. i implement formula in ...
0
votes
1answer
40 views

Is ROC or PR curve only the overall performance measure for classification

We can use ROC or PR curve to access the performance of the classifier,especially on imbalance data. But it is a curve with parameter threshold, even if we get a high ROC or PR performance, which ...
0
votes
1answer
35 views

How to get probability from the confidence score in SVM

In liblinear library we can get confidence score (the distance between decision hyperplane) in SVM solver for a binary classification problem, but if i want a probability value for membership in any ...
1
vote
1answer
15 views

Training lstm a sequence one item at a time

I am trying to train an lstm with a sequence and get the sequence classification for the whole sequence. I have sequences of varying length so I have one input neuron and I am feeding one item at a ...
2
votes
0answers
139 views

Huge Discrepancy between OOB and Cross Validation Random Forest

I am dealing with Random Forests at the moment. I observe huge discrepancies between OOB generalisation error estimation and cross validation. Originally, I used the scikit-learn package. But to ...
0
votes
0answers
12 views

How to use KL-divergence in naive bayes classifier to weight features?

I have a dataset consisting of 4 classes. I have implemented the Gaussian Naive Classifier (in Matlab). In the training phase I calculate the mean and variance for each feature and each class as well ...
2
votes
2answers
156 views

Robust softmax solutions for Theano?

I am implementing multilayer perceptrons with the softmax activation function over Theano. In some extreme cases I am running into problems with too high/low values in the softmax function that ...
-2
votes
0answers
11 views

Does theano support multi-machine version [on hold]

We use theano in the GPU in a single machine, I am just wondering if it supports multi-machine programming so that the speed will raise up?
3
votes
1answer
31 views

What's the measure to assess the binary classification accuracy for imbalance data

Now I have binary classification problem with positive samples roughly 100 times the number of negative samples. In this case the normal accuracy measure (predict == label) is not a good measure. What ...
2
votes
0answers
29 views

Finding idf only for text mining

We find tf-idf for training phase in text mining, however, in test phase, we need the tf for each element in test set, but should use idf in train set, so is there any api in python that can calculate ...
0
votes
0answers
11 views

Inversely proportional version of a nearest-neighbour results vector - how?

Short version: Given an input vector D of n values, what are the different methods that one can use to return a vector W such that each value in W is in inverse proportion to the magnitudes of the ...
1
vote
1answer
41 views

How do I implement missing value patterns?

I have a training data set and I was able to find some interesting patterns in the missing values, and I used binary variables in order to represent the missingness. I am going to train a model, say a ...
0
votes
0answers
21 views

tf-idf in text mining

I used sklearn of Python for getting tf-idf attribute in text analysis, but the problem is: I have about 78000 words in train_set, but the tf-idf matrix only has 39000 words. What is the problem ...
0
votes
2answers
163 views

Predict interesting articles: increase accuracy

I'm trying to write a GUI to display articles, and predict which articles I could like, based on the articles I previously liked. This post is the continuation of this one: ...