Methods and principles of building "computer systems that automatically improve with experience."
0
votes
0answers
54 views
How to identify a new pattern in a URL with a machine learning algorithm (Text mining)
I am trying to identify new patterns after analyzing a number of URLs. So let's say, I am investigating the hypothetical website Yoohle.com and their URLs have the following structure.
domain = ...
0
votes
0answers
20 views
How to create synthetic data from known weights
I'm doing some machine learning where I have lots of data and through optimization I'm trying to learn the weights for the model.
I'd like to check that my learning actually works correctly. For that ...
1
vote
1answer
47 views
How do I use Q-Learning to update values?
I am having trouble using the Q-learning algorithm to solve the problem in the link below.
http://math.stackexchange.com/questions/357397/how-do-i-use-q-learning-to-update-values
Any help would be ...
3
votes
1answer
68 views
PCA before train/test split
I have a dataset for which I have multiple sets of binary labels. For each set of labels, I train a classifier, evaluating it by cross-validation.
I want to reduce dimensionality using PCA. My ...
1
vote
0answers
52 views
Increasing the value of C in SVM (LibSVM) is not changing the accuracy at all
I am trying to learn SVM Classifier using some amount of training data.
and then I am predicting for another set (independent from training data)
I tried random C values from 0.000001 to 50000000 and ...
2
votes
2answers
80 views
ML with fastest classification speed
I have a data classification problem and I'm wondering what is the best machine learning approach to use for the particular constraints of my problem.
My constraints are as follows:
- the data ...
7
votes
1answer
270 views
Occam's razor obsolete?
I saw Vapnik's books about statistical learning... I read the first few chapters.
Anyway what surprised me the most was that he thought that the Occam's razor was obsolete.
I thought it was related ...
4
votes
2answers
59 views
Binary Classifier with training data for one label only
In some real-life problems such as authentication, we only have training data for one label (x is authenticated) while the other label doesn't have any data or only few entries (x is an imposter).
...
1
vote
0answers
32 views
Mixture model as a prior distribution
I've just started working with Bayesian models.
My question is in the context of hierarchical Bayesian model.
Suppose you have n models to train. However, some of these models are similar to each ...
2
votes
0answers
46 views
How should I distribute a classifier to customers?
When consulting, I often do my exploratory analysis and prototyping in R, and deliver results on the initial dataset to the client. The client wants to use the trained classifier in a production ...
0
votes
0answers
47 views
Normalizing SVM predications to [0,1]
I have trained an linear SVM which takes a pair of objects, computes features and is expected to learn a semantic similarity function between objects(we can say that it predicts whether the two ...
3
votes
3answers
165 views
How does PCA improve the accuracy of a predictive model?
I've seen in a kaggle challenge about digit recognition someone who used PCA before decision tree or other techniques.
I thought it was just for compressing data but he aimed to improve his score.
...
1
vote
1answer
53 views
Interpretation of Scree plots and Boruta Outcomes
I have 37 features in my dataset. I used Boruta package in R and according to its analysis, all the features are "important" and should be retained. I examined this result of Boruta and found that if ...
0
votes
0answers
40 views
Weighting and time series with machine learning
I'm trying to produce a model to predict the price of a product on the basis of several factors effecting previous time-stamped sales. I am certain that older sales are less relevant to the prediction ...
3
votes
5answers
128 views
How do you decide what your train, validation and test percentages are?
When splitting up my labeled data into training, validation and test sets, I have heard everything from 50/25/25 to 85/5/10. I am sure this depends on how you are going to use your model and how ...
0
votes
0answers
79 views
Modeling using neuralnet package in R - lots of issues
I am using neural net package in R. While I understand the basic neural network concepts, the details and back end is still a tough nut for me.
Currently all I can do is use brute force to change ...
1
vote
1answer
102 views
How can one show a Kmeans solution is unique?
Suppose we are given a distribution P and a constant K. We wish to minimize the kmeans objective w.r.t centers ${C1,..Ck}$:
What constraints on $P$ are known to imply that the optimal solution is ...
6
votes
1answer
152 views
Covariance matrix for Gaussian Process and Wishart distribution
I'm reading through this paper on Generalised Wishart Processes (GWP). The paper calculates the covariances between different random variables (following Gaussian Process) using squared exponential ...
1
vote
0answers
45 views
Fat-shattering dimension
A set of points $X = \{x\}$ is $\gamma$-shattered by a set of functions $\mathcal{F}$ if there are real numbers $r_x$ indexed by $x$ such that for any binary vector $b$ defining labeling of points ...
0
votes
1answer
47 views
Comparing two points corresponding to two different normal distributions
I have two multi variate normal distributions N1 and N2.
Say two points p1 is from N1 and p2 is from N2.
I want to get some statistical features from these two points. How can I do it?
I need a ...
2
votes
0answers
71 views
detecting circadian rhythm in a time series
I have a sensor that can detect minute changes in distance. It produces a time series.
I would like to point it at people and detect things like their sleeping pattern. How would one build a system ...
1
vote
0answers
34 views
Bayesian learning of tree distribution
this is my first post here.
I'm currently trying to compute the posterior predictive likelihood for a tree-structured distribution, following the paper Tractable Bayesian learning of tree belief ...
0
votes
2answers
75 views
Estimating probabilities using Bayes rule?
I am working on a past exam paper. I am given a data set as follows:
Hair {brown, red} = {B,R}, Height {tall, short} = {T,S} and Country {UK, Italy} = {U,I}
(B,T,U) (B,T,U) (B,T,I)
(R,T,U) (R,T,U) ...
2
votes
1answer
45 views
Predicting team performance
Are there any Machine Learning/ Datamining research paper which deals with predicting the performance of two team containing each $x$ players?
1
vote
0answers
39 views
Prediction with intervals as the independent variable
I have sample data that maps intervals to a number:
[3,7] => 1
[6,8] => 2
[6,13] => 3
[7,10] => 3
[10,13] => 4
The dependent variable's values ...
0
votes
0answers
39 views
Any basic Wavelet kernel implementation for SVM?
I have looked at at the toolbox by Alain from here http://asi.insa-rouen.fr/enseignants/~arakoto/toolbox/index.html but it comes with bad documentation. Does anyone have experience here implementing a ...
0
votes
0answers
42 views
Machine learning Jaakkola different sized data
I have an assignment for a machine learning problem and requires me to to compute $G=\{\|X_i-X_j\| \mid (X_i,Y_i),(X_j,Y_j)\, \text{that belong to training data}\, S\, \text{and} Y_i\neq Y_j\}$. In my ...
2
votes
2answers
111 views
Mutual Information based feature selection
Suppose I have a feature matrix $F = [f_1^T,f_2^T,...,f_m^T]$ whereby $f_j^T \in \mathbb R^{n_t \times 1}$ is the $j$th column vector of $F$ ($n_t$ is the number of different events/trials and $m$ is ...
0
votes
0answers
37 views
How to calculate entropy in log Scale?
I'm working on a problem , where a function returns the Log probability of P(X=x) . Now I would like to find the Entropy of X. But since the probabilities I get are log probabilities, again taking its ...
0
votes
0answers
91 views
Fitting a logistic regression using lassoglm in matlab
I am fitting a logistic regression model using lassoglm in matlab. I issued the following command
...
0
votes
0answers
45 views
Classifying high-dimensional data
I'm only learning about classification but why is it common practice to use PCA before using a Support Vector Machine?
Assuming I have 128*10 features and only 90 datapoints for each, do I need to ...
0
votes
0answers
9 views
Recall Precision curve, precision jumps at high recall [duplicate]
I am trying to plot a recall-precision curve for an object detection algorithm. In order to detect objects, I create a vote-map (2D histogram) in which object centres are voted for.
I then filter ...
0
votes
1answer
130 views
How to calculate recall-precision curves
I am trying to plot a recall-precision curve for an object detection algorithm. In order to detect objects, I create a vote-map (2D histogram) in which object centres are voted for.
I then filter ...
0
votes
1answer
122 views
understanding of libsvm output
I applied libsvm to build a text classifier. The output looks like as follows:
...
2
votes
1answer
133 views
Predicting Football match winners based only on previous data of same match
I'm a huge football(soccer) fan and interested in Machine Learning too. As a project for my ML course I'm trying to build a model that would predict the chance of winning for the home team, given the ...
0
votes
0answers
36 views
Which level of abstraction is appropriate when designing a Hidden Markov Model?
Bear with me, it is all new to me.
I have measured a thing repeatedly over a period of time and I have clustered the results. (Clustered the measured values without the time information.) The ...
0
votes
0answers
73 views
Wavelets and machine learning
I am trying to learn features from a signal using Wavelet transform and then apply ML techniques on it to classify a signal. The problem I am facing is that, at each of level of decomposition, my ...
3
votes
1answer
65 views
Variable importance randomForest negative values
I am asking myself if it is a good idea to remove those variables with a negative variable importance value ("%IncMSE") in a regression context. And if it gives me a better prediction? What do you ...
-1
votes
1answer
42 views
How does Logistic regression classifier modelize the dataset?
I'm working on a system that be able to detect the hand contour. So I have 270 instance in my dataset: 7 class of hand contour, 8 feature vectors of each instance.
Firstly, I used Weka to determine ...
0
votes
1answer
153 views
ROC curve and confusion matrix in classifier performance evaluation
I applied two different classifiers against the same validation set. It turns out that classifier A is better than classifier B in terms of ROC curve. However, classifier B is better than classifier ...
2
votes
0answers
121 views
Matrix factorization and gradient descent for recommender systems; user bias?
I've been reading about using Matrix Factorization techniques to do collaborative filtering. A popular thing to do seems to be to add user and item biases into the ratings prediction. What I don't ...
0
votes
0answers
27 views
Can very large numbers of samples throw off Naive Bayes?
I'm writing what is effectively a spam filtering system. I have about three million samples, mostly bag-of-words and top-tfidf based samples, and I'm seeing abysmal precision and recall. I'm trying to ...
2
votes
1answer
138 views
Logistic regression as classifier and overfitting
I am using logistic regression to classify data into two classes. The variable to predict (Y) is either 0 or 1.
I have found a rather good estimation of Y by logistic regression, and ended up using ...
0
votes
0answers
72 views
Description of all models in R's caret package? [closed]
I've been looking into machine learning recently, mostly using R. I just came across the caret package and it seems to be brilliant for quickly trying out different models. It seems a great tool for ...
1
vote
1answer
61 views
Derivation of the posterior over topics in LDA
When studying the latent Dirichlet allocation, I am not very clear about some procedures in their deriving equations. Please refer to the attached figure, how to understand those two steps, marked as ...
0
votes
2answers
43 views
difference in training and testing procedure of model
Can anyone please tell me the difference in training and testing of a model. I have developed 5/6 different single pass online learning algorithm (ets, ets+, evolving fuzzy modelling, SOFNN, ...
3
votes
1answer
77 views
Decision tree : handle attribute with many nominal values
I would like to build a decision tree from a training data. I have an attribute with many nominal values. For example, the department name attribute has about 20-30 values. I would like to group ...
0
votes
0answers
81 views
How to compute precision for a multiclass problem?
I have a question about calculating precision on a multiclass problem. If the true positives of some actual class is 0, and its false negatives is also 0, then how to calculate its recall? In this ...
0
votes
0answers
80 views
Predicting twitter activity using time series analysis
I'm interested to build a model for predicting how many tweets people I follow will probably tweet today (or by hour), based on their previous tweets in the last 60 days (or more).
Of course that the ...
1
vote
1answer
59 views
Non-independence of IVs in a random forest model
How is a random forest model affected if some of the variables are not independent?
