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

learn more… | top users | synonyms

1
vote
0answers
5 views

How do we get/define filters in convolutional neural networks?

How do i obtain filters from convulutional neural network(CNN)? My idea is something like this: Do random images of the input images (28x28) and get random patches (8x8). Then use autoencoders to ...
0
votes
0answers
1 views

How to account for different ratio of samples during training and detection using a support vector machine (svm)?

Consider the following object recognition case: Detection of objects in an image using a sliding window approach in combination with a svm model. During sliding window search using multiple scale ...
0
votes
0answers
5 views

How to find a good model for an object recognition case using a support vector machine (svm)?

Consider the following example of an object recognition case: I'm trying to detect objects in an image using histograms of oriented gradients (hog) features. The feature vector resulting from hog is ...
1
vote
0answers
20 views

Building a predictive model, regression with a long right tail

I am trying to build a, regressive, predictive model for a target time-series that is heavily skewed. You could think of the target as being like earthquake magnitudes or heavy rainfall. Most of the ...
1
vote
1answer
27 views

Does Deep network (e.g. # of hidden layer=2) always better than shallow network (i.e. # of hidden layer=1)?

I attempted to build a deep network (e.g. deep autoencoder) for some object classification, my result showed that the deep networks is worst than shallow network. However, from what I have read from ...
0
votes
0answers
13 views

Help explanation: Unreliable and uncertainty of prediction results (multiple runs)_detailed results included

Currently, I meet such questions when building Random Forest model using my data set. My full data set: X_lab: 839 * 469 and y_lab: 839 * 1 which is for all labelled data and X_unl: 20346 * 469 which ...
0
votes
0answers
8 views

Practical problem computing de k-nearest neighbors in CF?

I’m trying to apply de knn to a very dynamic system where users (like/dislike) items very frequently and new items became available all the time. My question is how often should the algorithm ...
0
votes
0answers
13 views

Regarding Naive Bayes and conditional independence

We all have been talking about how Naive Bayes may, in some cases, not perform well due to the fact that this assumes conditional independence of features and MOSTLY, this is not true for real world ...
1
vote
0answers
26 views

What is the relationship between vector space models & support vector machines?

Is there a relation between them? Specifically, if I have a VSM can I classify it through SVM?
0
votes
0answers
15 views

Graphically, how does the non-linear activation function project the input onto the classification space?

I am finding a very hard time to visualize how the activation function actually manages to classify non-linearly separable training data sets. Why does the activation function (e.g tanh function) ...
0
votes
0answers
7 views

LIBSVM for pre-computed kernel and zero bias (b values is zero)

I want to do binary classification and I'm using LIBSVM library for that. I have a precomputed Kernel and my bias value (b) is zero. Can I do this in LIBSVM or do I have to use some other library? ...
0
votes
0answers
27 views

Visualize a large binary matrix with instances and three classes

I have a matrix with 9500 columns and 1000 rows. Each row represents an instance. I have three classes and an instance belongs to a class. Each column represents a binary feature. That is, each cell ...
0
votes
0answers
8 views

Meta-parameter search for elastic net regularization of general objective function

In their 2004 paper on elastic net regularization, Zou and Hastie present an efficient method for finding the meta-parameters by folding the $L_2$-regularization component into the OLS problem and ...
0
votes
0answers
19 views

Does Text Classification using SVM needs to use a dictionary to build the features vectors?

I'm kinda new to this, but I want to do an experiment I need your guys help. Open to all suggestions. Let's say I have around 5000 user accounts for which I only have several attributes [first name, ...
0
votes
0answers
11 views

algorithm to predict cost function

The goal of problem is to predict the weight for missing data . I have a dataset of categorical type as shown below ...
0
votes
2answers
40 views

Different results from several “passes” of Random Forest on same dataset

I've been playing around with the German Credit dataset available in Kuhn & Johnson's caret package for ...
1
vote
1answer
23 views

How to apply the output layer function in a neural network

I am implementing a Neural Network in a somewhat different fashion. I train my neural network locally using a small subset, and export the weights. My goal is to test the neural network in a ...
0
votes
0answers
32 views

Stochastic Gradient Descent for Logistic Regression always returns a cost of Inf and weight vector never gets any closer

I am trying to implement a logistic regression solver in MATLAB and i am finding the weights by stochastic gradient descent. I am running into a problem where my data seems to produce an infinite ...
0
votes
1answer
40 views

What is the convex hull in ROC curve?

I'm reading a paper about ROC and PR curves. They mentioned the ROC convex hull but they don't define it or say what it is. Can someone please tell me the meaning of it? What is a convex hull in ROC ...
2
votes
2answers
40 views

An intuitive meaning of the area under the PR curve?

Wikipedia says that an interpretation of the area under the ROC curve is: "the area under the curve is equal to the probability that a classifier will rank a randomly chosen positive instance higher ...
4
votes
2answers
191 views

How to do exploratory data analysis to choose appropriate machine learning algorithm

We are studying machine learning via Machine Learning: A Probabilistic Perspective (Kevin Murphy). While the text explains the theoretical foundation of each algorithm, it rarely says in which case ...
0
votes
0answers
12 views

Usage of libsvm with RBF kernel and no Offset

I'm using libsvm for the binary classification and using a precomputed Kernel. In my particular problem there is no bias term (it's zero). Is there anyway to adjust the bias term in libsvm (and not ...
-1
votes
1answer
26 views

Apriori Algorithm Support Calculation [on hold]

Why Apriori functions in arules library in R returns different values that what it should be? The sample data in the pic below. Each line between {} means that it is a single transaction. So ...
0
votes
0answers
11 views

Why does Co-ordinate descent work? [closed]

If it works does it mean that "function is convex if it's convex in any direction"?
2
votes
0answers
21 views

R programming, correlation of quantitave variables with one qualitative variable

I have a flat CSV file that has one column of student names, one column of grades (outcomes) coded as a factor A-F, and about 100 columns of test scores (independent variables) of various sorts, on ...
2
votes
2answers
135 views

Gaussian noise model derivation

I have the following linear regression model, $y = f(x;w) + n$, where $y$ is the vector of true labels, $x$ is the observed data, $f(x;w) = w^Tx$, and $n$ ~ $N(0, \sigma^2)$ is the noise. Why then ...
0
votes
0answers
40 views

The activities of the machine learning scientists at Amazon [closed]

I'm looking for a job in machine learning and data mining and I came across Amazon Germany where they have offers as machine learning scientist. Well, I don't know if I fit for that since I'm a master ...
1
vote
0answers
17 views

Using standard machine learning tools on left-censored data

I'm developing a forecasting application whose purpose is to allow an importer to forecast demand for its products from its customer network of distributors. Sales figures are a pretty good proxy for ...
1
vote
2answers
26 views

What does “permutation invariant” mean?

I have seen a term "permutation invariant" version of the MNIST digit recognition task. What does it mean?
1
vote
2answers
52 views

What're the differences between PCA and autoencoder?

Both PCA and autoencoder can do demension reduction, so what are the difference between them? In what situation I should use one over another?
1
vote
1answer
38 views

I am having trouble understanding (and implementing) logistic regression for classifying into three classes

(For reference, i am using Kevin P Murphy's Book "Machine Learning: A Probabilistic Perspective" and implementing with MATLAN - without any toolboxes) I have a dataset with 392 samples (rows), each ...
0
votes
0answers
19 views

How to compute F-statistics for each features of regression models in glmnet?

I have learned lot's of Lasso regression models(20000) using glmnet. I need to compute somehow test statistics for each features of models. like F-statistics,... Can I do this using bootstrapping ? ...
0
votes
0answers
21 views

How to improve linear model generalization when autocorrelation is present?

I have features $X_t$ and response $Y_t$ (all continuous variables) and my objective is to find the best estimate of $f(X_t)=Y_t$ where $f$ is linear, and 'best' is defined as lowest generalisation ...
2
votes
1answer
22 views

What does pre-training mean in deep autoencoder?

I am confused by the term "pre-training". What does it mean in deep autoencoder? And how does it help improving the performance of autoencoder? (I know this term comes from Hinton 2006's paper: ...
0
votes
1answer
16 views

What is the toolkit that implements Cost sensitive Support Vector Machine?

I need implementation of cost sensitive support vector machine. The cost is different for each training example (unlike each class). So problem is to solve $max_\alpha$ $-1/2 \sum_{i,j} ...
1
vote
0answers
14 views

How to share weights of between regression models when you learning them simultaneously?

I have many phenomenons which I want to model them as lasso regression problem. every phenomenon have it's own distinct features set. but for some phenomenons, the subset of features set are the ...
0
votes
0answers
18 views

Convolutional neural networks: Aren't the central neurons over-represented in the output?

[This question was also posed at stack overflow] The question in short I'm studying convolutional neural networks, and I believe that these networks do not treat every input neuron ...
0
votes
0answers
16 views

Multiple Class Logistic Regression Coefficents unstable when classes well separated

The following is a quote about the performance of Logistic Regression on multiple classes when the individual classes are well separated by the book ISLR: "When the classes are well-separated, the ...
0
votes
0answers
19 views

Why would a reasonable range of the regularization parameters $\lambda$ be up to the maximum eigenvalue of the kernel matrix?

I was wondering, how do you choose a reasonable range for the regularization parameter $\lambda$ for regularized least squares when doing k-fold cross validation? I was told that a reasonable range ...
0
votes
0answers
10 views

Binary classification with KNN

I post here because I don't know how to improve the performance of my binary KNN. The problem is that I have 99.8% Specificity and only 82% Sensitivity, but I'd rather have more Sensitivity than ...
2
votes
0answers
39 views

How to predict the winner in a set?

At the moment I am dealing with some kind of supervised regression problem, which differs from the stuff I have found on the internet. So here's what the data looks like. (I am working with R) I ...
2
votes
1answer
39 views

Support vector regression versus kernel ridge regression

I have a question concerning the difference between support vector regression and kernel regression. I will try to write down all the math so no misunderstandings arise (hopefully). Let's begin with ...
0
votes
0answers
19 views

Do we need to do sigmoid after pooling in convolutional neural network?

I understand we need to do sigmoid transformation after convolution step in building convolutional neural network(CNN). Do we need that also after pooling step? Like: Let assume: CL = convolution ...
0
votes
0answers
25 views

What is the correct architecture for convolutional neural network?

I have seen several different architectures for convolutional neural network (CNN). I am confused which one is the standard and how do I decide what to use. I am not confused by the number of layers ...
2
votes
1answer
42 views

Maximum Degree of Polynomial Regression

If we have 100 data points and want to perform polynomial regression, the maximum degree of our polynomial is n-1, where n is the number of data points. In this case, the maximum degree would be 99. I ...
2
votes
1answer
71 views

What is Bayes decision rule?

Assume binary classification i.e. $y \in \{-1,1\}$ and that the underlying joint probability distribution generating the data is known i.e. $P_{x,y}(x,y)$ is known I was told that Bayes decision ...
3
votes
4answers
77 views

Sample selection algorithms to ensure that training & validation sets are representative

Currently, I am encountering a question, which is how to selection representative samples (training set and test set, even validation set) from the whole data set? I would like build a classification ...
0
votes
0answers
32 views

Change in objective function optimization - Regularization in Logistic Regression

If I have the objective function of Logistic Regression to optimize by maximizing it, would it change to a minimization problem when I add regularization term to it? Or can I still solve the ...
0
votes
1answer
29 views

In Convolutional neural network, what does fully-connected layer mean?

There are convolution layers, pooling layers, and possibly a classifier layer (e.g. softmax layer) in convolutional neural network(CNN). I heard that there is also a fully-connected layer, what is ...
0
votes
1answer
64 views

Generating PMML export of a gbm model in R?

Is it possible to generate PMML of a gbm model? When I try to use the pmml library, I get an error: Error in UseMethod("pmml") : no applicable method for 'pmml' applied to an object of class ...