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

learn more… | top users | synonyms

0
votes
1answer
30 views

Grid Search for hyperparameter and feature selection

So I need to select my hyperparameters and also my features. A full grid search of the space of hyperparameters and features is too computationally intensive, so what I am doing instead is for each ...
1
vote
0answers
24 views

How to determine appropriate number of features and also which features to select?

So I have a dataset which I am using K fold cross validation on to select the number of features and which features should be selected. As I understand it, I would set the number of features to be ...
0
votes
0answers
26 views

How to build feature vectors from profile data

I want to build feature vectors from data of my test set, which contains profiles of people. I always want to compare two profiles to each other. Thus my features are: - Same surname ∈ {undefined, ...
1
vote
1answer
27 views

Ploting classifiers' performance results for different datasets

I have five different data sets, and I applied five classifiers. I have also five performance evaluation metrics: precision, recall, F-measure, Accuracy, and AU-ROC. Is there any good plot that I can ...
0
votes
1answer
41 views

Issue in training Hopfield network and convergence problem

I am learning how to use Hopfield Neural network as a context addressable memory. The objective is to obtain a fixed point of the network which indicates an equilibrium state. This state vector ...
3
votes
1answer
81 views

Machine learning techniques for spam detection, and in general for text classification

I am going to configure a system for spam detection. What I have is a dataset of labeled (spam/not-spam) strings containing, mostly, sentences. I have a background in machine learning techniques, but ...
2
votes
0answers
31 views

Why are after training 5-state HMM only few entries of transition matrix left greater than zero?

I try to create the speech recognition system based on 5-state HMM + Multivariate Gaussian function. I use my own feature vector derived from MFCC (Mel-frequency cepstral coefficients). The problem is ...
0
votes
0answers
16 views

Posterior distribution for LDA and Newdata

I am using the 'topicmodels' package in R. I tested the posteriori probability for newdata over jss_LDA result by this code : ...
0
votes
1answer
26 views

One vs All and One vs one in svm?

What is the different between onee vs all and one vs one SVM classifier?? Is One vs All mean = 1 classifier to classify all types /categories of the new image and one vs one mean= each type /category ...
1
vote
1answer
114 views

When is a proper scoring rule a better estimate of generalization in a classification setting?

A typical approach to solving a classification problem is to identify a class of candidate models, and then perform model selection using some procedure like cross validation. Typically one selects ...
-1
votes
1answer
19 views

Converting FuzzykMeans to SphericalFuzzyKMeans?

I grabbed an implementation of FuzzyKMeans (FuzzyCMeans) from the nightly build of the Apache Commons Math library, but I now realize I need to use Cosine Similarity instead of the Euclidean Distance. ...
0
votes
0answers
17 views

Q-Learning: Can you move backwards?

I'm looking over a sample exam and there is a question on Q-learning, I have included it below. In the 3rd step, how come the action taken is 'right' rather than 'up' (back to A2). It appears the Q ...
0
votes
1answer
39 views

Automatic labeling of training set

I have once meet the following question, given a training set, is that possible to do the automatic labelling? In addition, if this training set consists of plain text files, is that possible to know ...
2
votes
1answer
34 views

In decision tree construction, can a good splitter have low information gain?

I have a data set with a candidate splitter variable that is a natural choice from the business perspective. It has two values, and the distributions of the target when conditioned on the two values ...
1
vote
1answer
54 views

How to interpret the number of k in k-nearest-neighbour classifier?

I have done some classification work using a k-nearest-neighbour classifier (kNN). And the classification performance is evaluated using cross-validation method. Some testing code from Matlab Help are ...
3
votes
2answers
315 views

How can stochastic gradient descent avoid the problem of a local minimum?

I know that stochastic gradient descent has random behavior, but I don't know why. Is there any explanation about this?
0
votes
0answers
18 views

Cross validation training support vector machine

Can somebody explain me how to perform cross validation while training Support vector machine via some example?
1
vote
0answers
27 views

What is the meaning of the term “enrichment” when performing cross-validation?

Trying to understand a discussion of a 5-fold cross-validation process to validate a predictive model and its results, there is a particular phrase which has me stumped, i.e.: The predictions of ...
1
vote
2answers
50 views

Area under the ROC curve or area under the PR curve for imbalanced data?

I have some doubts about which performance measure to use, area under the ROC curve (TPR as a function of FPR) or area under the precision-recall curve (precision as a function of recall). My data is ...
0
votes
0answers
10 views

Dictionary matrix for online sparse coding doesn't learn [migrated]

Following Online Dictionary Learning for Sparse Coding and using an inference function I've found on the Stanford's website, I'm trying to online learn a Dictionary matrix. Tested both on resized ...
2
votes
2answers
129 views

The difference of kernels in SVM?

Can someone please tell me the difference between the kernels in SVM: Linear Polynomial Gaussian (RBF) Sigmoid Because as we know that kernel is used to mapped our input space into high ...
1
vote
1answer
52 views

What kind of model is used by 20 Questions?

Which kind of machine learning concept / model is used in 20 Questions? Is this kind of thing best solved by a neural network? Where I can read something about it?
5
votes
3answers
237 views

Why is AUC higher for a classifier that is less accurate than for one that is more accurate?

I have two classifiers A: naive Bayesian network B: tree (singly-connected) Bayesian network In terms of accuracy and other measures, A performs comparatively worse than B. However, when I use the ...
3
votes
1answer
63 views

SVM basic theory?

I have some questions about SVM: In SVM there is a nonlinear and linear SVM. What is the difference between them? To do classification in SVM, we will find the linearly separable boundary ...
5
votes
1answer
38 views

Is the value of $\alpha$ the same for all support vectors (SV) in the dual and what is the reason for it if they do or don't?

Consider the dual with no offset and not slack. In the dual we have that for data points that are support vectors: $$\alpha_t > 0$$ and that the constraint is satisfied with equality (since a ...
1
vote
1answer
24 views

Different formulations for SVM with slack variables (primal)

I have seen two different ways to formulate the SVM optimization but I was not sure what the difference was between them or if there was any difference. First formulation: $$min ...
1
vote
0answers
28 views

Estimating the covariance matrix in LDA

I was trying to derive the equations from page 109 in "elements of statistical learning" (image below) To be honest, I am not sure how the covariance $\Sigma$ is estimated (the third bullet point in ...
2
votes
1answer
116 views

Gaussian Process Kernel and Ridge Regression

Can a Dual Ridge Regression produce the same prediction results as a Gaussian Process with a polynomial kernel $K(x,x')=(x^Tx'+1)^2$ in less time complexity (GP is $O(n^3)$ ) using Cholesky ...
0
votes
0answers
13 views

Imbalanced values in the feature set of training and testing samples in SVM (Multi class classification)

Currently I only know about the imbalanced in the structure of data set (e.g. too many positive samples, few negative samples..). But how about imbalanced in the value of features in each samples? For ...
3
votes
1answer
60 views

SVM primal formulation: does the constants constraint matter?

When finding the maximum margin separator in the primal form we have the quadratic program $$min\frac{1}{2}||\theta||^2$$ $$\text{ subject to: } y^{(t)}(\theta \cdot x^{(t)} + \theta_0) \geq 1, \ ...
0
votes
0answers
15 views

Why even my training data failed during the prediction of libsvm [duplicate]

Currently I'm using libsvm for my one class classification problem. I have 10 samples in my training set, 5 samples in my testing set, both of my training and testing set is scaled by svm_scale, then ...
0
votes
0answers
19 views

Classifier jars for java

I have a train file which has categorical features like IN JJ PRP_VBP VB NN PRP$ . . . The third column is the ground truth and can have value only out of ...
0
votes
0answers
35 views

Neural Network, dependence among outputs?

Is there a way to train a neural network in the following manner: You have $n$ observations in the training set. The neural net will start with random weights, and produce $n$ outputs. I want to ...
1
vote
1answer
115 views

How to design neural networks for pattern recognition in biometry?

Having read numerous texts regarding neural networks and their characteristics, I am getting increasingly confused, paradoxically – I am looking for a brief explanation or references to the right ...
1
vote
1answer
60 views

How to define train and test sets in financial time series for estimating machine learning parameters

After reading some material, I found few options for defining train and test sets: Just splitting with no change. Accumulating/moving window of train set. Leave a relatively small (warming) period ...
0
votes
1answer
84 views

What is SVM regression? Is it for regression or classification?

I'm trying to understand what is SVM regression. It's used for classification or regression? Can someone give an intuitive understanding of it?
0
votes
0answers
28 views

Best Methods for data-mining neuroimaging data with 1000 subjects

I am part of a team tasked with performing exploratory analysis of a large data set containing neuro-imaging scans. For each scan I will likely calculate some variable that relates to brain function - ...
0
votes
0answers
23 views

Significance test for multiclass classifier

In a multiclass classification problem, I want to measure the significance of my classifier against the null hypothesis (in this case, chance level). In this paper, in section 3.4, for a binary ...
1
vote
0answers
20 views

What do I call a set of datasets

For training a Machine learning model, I have 3 datasets: Training Validation Testing Normally I obtain there by dividing up the full dataset into pieces. I've created a function in my code to ...
0
votes
0answers
14 views

Choosing the values of a proper subset of features to maximise regression tree output

Suppose I have a regression tree and feature set $X$. Suppose that the feature set is composed of $X:=\{X_0,X_1,...,X_{100}\}$, where each $X_i \sim N(0,\sigma^2)$. Suppose that ...
2
votes
1answer
40 views

Dimension of weight vectors in SVMs

For a given set of features (say with dimension a) and for a given set of labels (say m labels), how to relate the given features with the weight vector of the SVM in general? Will it be equal to ...
3
votes
3answers
132 views

Multiple Regression and number of parameters to include for a learning algorithm

I am quite new to Machine Learning and come from a computing background. I have a quite big set of features (~50) with about 4k observations. Is it correct thinking to include all of them in a ...
1
vote
2answers
76 views

What's the difference between kernel and stepwise logistic regression?

I am confused by different terms of logistic regression. What are the differences between stepwise, kernel, forward, and backward?
1
vote
0answers
41 views

Exploratory Analysis - finding the most important factor

I have a dataset of 113 variables. In exploratory analysis the first thing I want to know is what are the most important factors on a single variable (revenue). I learned that naive Bayes would ...
-1
votes
1answer
45 views

Clustering algorithms assigning probability values

I have a distance matrix for some data I want to cluster. However, I don't just want to assign elements to clusters, but I also want to assign a probability for each element to belong to each cluster. ...
1
vote
1answer
61 views

building a classification model for strictly binary data

i have a data set that is strictly binary. each variable's set of values is in the domain: true, false. the "special" property of this data set is that an overwhelming majority of the values are ...
1
vote
1answer
64 views

Is it abnormal that out-of-sample fit is better than in-sample?

I'm using Eureqa, as machine learning tool to fit a formula to my data. I found out that the formula fits my test data better than my training data! Is this abnormal?
2
votes
1answer
149 views

'Uniformization'?

I am looking for a better term for what I call 'uniformification', where I change data to make it more close to uniformly distributed. I am doing a project in which I try to make the output of a ...
0
votes
0answers
17 views

How to estimate False Discovery Rate from p-value distribution?

I have learned many models and I calculated p-values for the cross-validation errors. I want to select significant models based on the false discovery rate (FDR). How can I estimate the FDR from ...