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

learn more… | top users | synonyms (1)

0
votes
0answers
8 views

how to choose parameters for paralysis of neural network classificator?

I have multilayer feedforward neural network and I learn it to do multi classification using back propagation method. Can you provide an example of how to create paralysis of neural network (when nn ...
10
votes
1answer
178 views

Computation of the marginal likelihood from MCMC samples

This is a recurring question (see this post, this post and this post), but I have a different spin. Suppose I have a bunch of samples from a generic MCMC sampler. For each sample $\theta$, I know the ...
0
votes
0answers
10 views

How to classify group of events (or actions) from an event log

Given data that contains events carried by group members (e.g., in the format of 'group_id, member_id, event_id, timestamp'), and label for such history data mapping a sequence of group interaction ...
1
vote
0answers
33 views

Repeated training examples in Gradient Descent

I am new to machine learning and trying to understand stochastic gradient descent. I understand in stochastic gradient descent, in each epoch, randomly an example is picked and given to the model. So ...
1
vote
1answer
37 views

Why are gradient boosting regression trees good candidates for ranking problems?

I have been reading up on gradient boosting machines, and in particular GBRT's. I've come across numerous mentions (and finally tracked down some papers) on applying these models to ranking problems - ...
1
vote
0answers
27 views

Can a model recognize the interactions between variables?

I have a data set similar to the following one, ...
-1
votes
0answers
15 views

handwritten hindi numeral recognition using SVM

I have used LIBSVM to classify the hindi handwritten numerals . I have got a result of 93.38 % , when I used C=500 and Gamma = .004. I would like to get an idea about k fold cross validation and the ...
0
votes
0answers
7 views

Find a maximum value for input data

I have a set of 20x20000 input variables and 1x20000 target values. I want to find a best combination of 5 variables out of 20 on a condition that other 15 variables are constant to get the best ...
0
votes
0answers
14 views

Method to determine whether or not users had a bad experience based on multiple variables: Average Bandwidth, Latency, and frame rate

I would like a recommendation on the best statistical method to use, as well as any suggested R packages to achieve this goal. I have three variables, Bandwidth, Latency, and frame rate for a set of ...
0
votes
0answers
3 views

forward wrapper feature selection method on python [migrated]

I've coded a decision tree for my dataset. I'm using the semeion.data which is found at https://archive.ics.uci.edu/ml/datasets/Semeion+Handwritten+Digit. Now I need to find out how to use ...
0
votes
0answers
18 views

Can supervised machine learning techniques infer the formula (if it exists) for a statistical model?

Suppose that we have response data $y_i$ generated by a specific mathematical function $y_i=\mathcal{F}(X_i)+e_i$ where $X_i$ is a vector of predictor variables with random error term $e_i$. Without ...
3
votes
1answer
51 views

Bayesian treatment of outliers

In a supervised learning problem, I have a training dataset $D$ comprised of samples $x$ and their corresponding labels $\omega$. From this data, I attempt to learn the true distributions ...
1
vote
1answer
21 views

The unique variance in Factor analysis

This might be a rather simple question, but this is troubling me a lot? Please help. Suppose in a factor analysis model, there are three variables $x_1$, $x_2$ and $x_3$, and two latent factors ...
0
votes
1answer
18 views

Handling missing/rare levels in predictor in data samples

Let us assume we have a dataset with one catigorical variable, which is represented in R as a factor. I am performing crossvalidation to assess models, for which I need to perform stratified sampling ...
0
votes
1answer
48 views

Random Forest model good train and test performance but bad “real world” performance

I am working on a classification problem where I need to classify objects based on a visual data. There are a couple hundred different classifications to be made and I have around a million plus ...
0
votes
0answers
37 views

How to know the importance of all the variables and levels in them using Random Uniform Forest in R?

I have a dataset containing 3 parameters (Region( factors - say US,UK,Aus,NZ),Domain or Industry( factors - say IT,Electrical,Mechanical) and Scope - good or bad). Using Random Uniform Forest package ...
0
votes
0answers
9 views

Classification of overlapping hetegenerous cell nuclei

We are two people doing a image analysis project on segmentation of cell nuclei. Our data set consist of about 300-400 cell nuclei, from 10-15 images containing different cell types. Our main problem ...
0
votes
0answers
10 views

How to deal with heterogeneous data set

Problem I have a dataset that contains three types of objects: simple objects, object groups and meta-groups. Meta-groups contain simple objects and groups. Each simple object correspond to an ...
0
votes
0answers
17 views

Best approach for calculating confidence score (p-value) - for Bayes models

This is a generic question on the best approach for giving a p-value to a "most likely" solution given by a graphical Bayes model. I am building a graphical model in Stan, for finding the most likely ...
1
vote
1answer
20 views

Can feature with high positive correlation have opposite weights sign? [duplicate]

I have two features for a binary classification problem which are highly positively correlated. (0.79) But when I build a logistic regression classification then I see their weights are opposite in ...
0
votes
1answer
29 views

How and what do I train in my Convolutional Neural Network [closed]

I have been trying to research and implement a convolution neural network in c++, and I think I understand the basic architecture of it. My problem is that I am incredibly confused as to what is ...
1
vote
0answers
22 views

Flexible prediction with neural network or other method

I want to use neural network for my first time, but I need to check if it fits for my case. So, my idea is to teach a model on data like Y = f(X1, X2...XN) and ...
0
votes
1answer
18 views

Large number of positive labels in classifier when actual population has few

I have been tasked to help with a sort of classifier. In the make up of the problem the set we want to identify as "Positive" is know to be very very small. However the training set I have been given ...
1
vote
1answer
34 views

deep learning in mobile apps [closed]

Can deep learning be applied in mobile apps, and if possible how? Is it possible ? to deep learning needs more computational cost? how can we minimize the computational cost for deep learning on ...
3
votes
1answer
51 views

What's an intuitive explanation for why MAP is variant under parameterization?

I understand why MAP is variant under parameterization mathematically, but I don't really understand it intuitively. To help me out, my professor gave me an example where reparameterizing MAP ...
0
votes
0answers
9 views

How does Weka chiSquaredAttributeEval generates single attribute selection list while Chi Square itself is class based?

I have implemented my own Chi-Square ranker in C# however the example i found on the internet shows that Chi-Square ranks the each attribute within its class However Weka generates attributes as a ...
1
vote
1answer
37 views

Attribute weighted KMeans

I'm trying to use Kmeans clustering, with an intent to find out clusters by weighting the attributes. Eg. if attribute A matters less than attribute B then the output should put more weight on ...
0
votes
1answer
66 views

What model to use for predicting future expenses of an individual?

I am currently working on a Personal Finance application, which tracks expenses of a person. When entering an expense entry, the user selects the category of the transaction (e.g. 'Bills', 'Food', ...
0
votes
0answers
4 views

Activation of different states in a neural network

hello i want to ask if it is possible to create a neural network in mat lab that is able to have at least 4 states (by states mean something happening parallely not the output )which gets activated ...
0
votes
0answers
18 views

Queries regarding 'logistftest' function of logistf package for model comparison

I am a user of r package ‘logistf’, I have 2 queries regarding the ‘logistftest’ function when doing model comparisons, I would appreciate if anyone can give helps. I am comparing two models (one ...
0
votes
0answers
33 views

Are there better algorithms than Convolutional Neural Networks?

I was wondering if there are any better algorithm/paradigm than Convolutional Neural Networks for tasks related to images ? Whichever problem domain I have picked up (Image classification, scene ...
0
votes
0answers
32 views

Understanding LDA inference

It is said that the key inferential problem that needs to be solved to use LDA (latent dirichlet allocation) is that of computing the posterior distribution $p(\theta,z | w, \alpha ,\beta)$. I know ...
0
votes
1answer
37 views

Neural Net Matrix Multiplication

I'm trying to figure out the matrix multiplications for the implementation of a single hidden layer neural net for MNIST digit recognition in Python. Like the following: ...
0
votes
0answers
11 views

Auto-association with backpropagation and hopfield network

When I was studying back propagation the professor mentioned that we could use auto associative memory to extract some information, but it seems that the Hopfield network (recurrent network) is better ...
2
votes
1answer
38 views

Word-topic matrix in Latent Dirichlet Allocation

In the Latend Dirichllet Allocation model described in Wikipedia (https://en.wikipedia.org/wiki/Latent_Dirichlet_allocation), is $\beta$ the word-topic matrix? I understand that $\beta$ is the ...
3
votes
1answer
39 views

Selecting ARIMA Order using Rolling Forecast

I'm wondering if a rolling forecast technique like the ones mentioned in Rob Hyndman's blogs, and the example below, could be used to select the order for an ARIMA model? In the examples I've looked ...
0
votes
0answers
20 views

Difference between rulefit and random forest

I'm trying to understand the difference between these a bit better. I understand pretty well how random forests work but I guess I'm more hazy on rulefit and how exactly it's different. I know rulefit ...
1
vote
2answers
74 views

One-vs-many/One-vs-all - what value to use as probability?

I have constructed SVMs to do a one-vs-many approach to classification. Let's say I have 3 classes and I train 3 SVMs in a one-vs-many format. This gives me 3 SVMs each trained positively on one of a ...
1
vote
0answers
15 views

How effective is connectionist temporal classification on a small dataset of 30 hour audio? [closed]

I am performing speech recognition on my local language and this seemed to be a scalable solution. Please provide suggestions
0
votes
0answers
19 views

Limited-memory BFGS and MaxEnt Models

According to Wikipedia, "L-BFGS has been called 'the algorithm of choice' for fitting log-linear (MaxEnt) models", however I'm having trouble finding a good reference to show how incorporating the ...
0
votes
1answer
28 views

Feature selection step before decision tree?

I want to use rpart (a R package) to build a decision tree model. The data is a high-dimensional expression matrix, with ~50,000 predictors and ~500 samples. The response is a categorical variable. ...
1
vote
0answers
9 views

What's the value of f1 score when true positives are zero? [duplicate]

f1 score is defined as F1 = 2 * (precision * recall) / (precision + recall) However if true positives are zero, this gives 0/0. How to deal with this scenario?
0
votes
0answers
13 views

Finding a number that considers %worse outcomes and %better outcomes

I have a baseline algorithm and lots of test algorithms. The baseline algorithm's performance is compared to every other algorithm. What I'm left with is a table like so: ...
0
votes
0answers
15 views

Comparing algorithms: runtime vs accuracy

I am comparing several algorithms with respect to their runtime and accuracy on several datasets. I follow [1] and compute the average rank using the accuracy and apply the Friedman and Nemenyi tests. ...
3
votes
1answer
71 views

What is principal subspace in probabilistic PCA?

if $X$ is observed data matrix and $Y$ is latent variable then $$X=WY+\mu+\epsilon$$ Where $\mu$ is the mean of observed data, and $\epsilon$ is the Gaussian error/noise in data, and $W$ is called ...
0
votes
0answers
19 views

Classification on sequential data

Context: I am working on a classification project where I recommend items to customers based on their past purchase history. Question: How will "time leakage" affect training? Example: Let's say ...
1
vote
1answer
34 views

Machine learning step order question

I have been working on this project for over a year now and I believe i finally have things figured out. Mainly i'm looking for any suggestions or things i'm doing wrong with my process, but i also ...
0
votes
0answers
37 views

A question on conditional gaussian distribution

The book on Pattern Recognition (by Bishop) begins the section on conditional gaussian by saying: An important property of the multivariate Gaussian distribution is that if two sets of variables ...
3
votes
2answers
54 views

In machine learning, Is it better to have class ratios balanced or representative of the population?

In the context of machine learning, let's say you have a problem in which classes in the real population are not balanced - eg Class A occurs 80% of the time and Class B occurs 20% of the time. In ...
0
votes
1answer
14 views

Machine learning algorithm for predicting 'action' performances

I'm given: the current state of the system S, for which the objective function C can be calculated. There's also a set of possible actions a1, a2, .... . Each action leads to a change in the state. ...