Stack Exchange Network

Stack Exchange network consists of 175 Q&A communities including Stack Overflow, the largest, most trusted online community for developers to learn, share their knowledge, and build their careers.

Visit Stack Exchange

Questions tagged [machine-learning]

Machine learning algorithms build a model of the training data. The term "machine learning" is vaguely defined; it includes what is also called statistical learning, reinforcement learning, unsupervised learning, etc. ALWAYS ADD A MORE SPECIFIC TAG.

0
votes
0answers
18 views

Scikit-Learn SVC Porbability Function

I use scikit-learn to train a SVC with 'poly'-Kernel and propability-paramter enabled. Most of the time the prediction and the probability assigned to the prediction is correct. That means: ...
0
votes
0answers
8 views

Pairwise matching of case and control group for a machine learning classifier

If I want to test whether an illness is associated with alterations of a dependent variable Y (example: grey matter volume) I can perform under some assumptions a t-test. If I am aware of a confounder ...
0
votes
0answers
13 views

Conditional model with correlation between estimations

I am trying to estimate Click-Through-Rate (CTR) given the following two models: $$P(Click|Visible)$$ $$P(Visible)$$ The output is: $$P(Click) = P(Click|Visible)*P(Visible)$$ Unfortunately, $$P(...
1
vote
1answer
50 views

Encoding of categorical variables with high cardinality

For unsupervised anomaly detection / fraud analytics on credit card data (where I don't have labeled fraudulent cases), there are a lot of variables to consider. The data is of mixed type with ...
0
votes
0answers
36 views

Distribution to model Binomial distribution with parameter p in trial n dependent on result from trial n-1?

I'm wondering how one can model a Binomial distribution as described in question. E.g., p = 0.5 for trial n = 0; p(n+1) = p(n) + 0.01 if for trial n Bernoulli(p(n)) samples to 1, else p(n+1) = p(n) - ...
1
vote
2answers
39 views

Does the MSE values of regression coefficients sum up to the MSE value of the regression model in which the regression coefficients are included?

I think either i dont understand something or i try to mix something that are different things. The mse value of a regression coefficients tells me how good i estimated the coefficent. Does it mean ...
0
votes
0answers
15 views

How to predict the complete Ranking (1-6) of six Machines. (Multi target prediction)

I would like to predict the "Ranking" (1-6) of all 6 coffeemachines with the Big Five personality traits (OCEAN) and i don't really know which Algorythm would be the best for this task. Any ...
0
votes
1answer
49 views

Learning rate constant in Random Forrest calculation

I saw the following in a Random Forrest calculation. My understanding of logarithms is not intuitive, I always have to look them up. It was asked: ...
1
vote
0answers
9 views

Deriving Multiplicative Update Rules for Regularized NMF

After reading the following CrossValidated post, I cannot derived the correct multiplicative rules for regularized NMF from this paper. They obtain the coefficients $|I_u|$ and $|U_i|$ in the ...
0
votes
0answers
13 views

Stationarization of 2-dimensional Time-Series

I'm trying to perform a Gaussian Process Regression on time-varying data of the form (t, x, y, z), where t is the time when ...
0
votes
0answers
39 views

Hyper-parameter tuning of neural network: do we always need it? and how to do it efficiently? [on hold]

I am studying the effect of using transfer learning (pre-train a part of network, freeze it, and use as part of new network to bootstrap the learning of a new task) of vs training a neural network ...
1
vote
0answers
11 views

What does “shot” “way” “task” mean for few shot learning? under RL setting as well?

Recently got into few shot learning, and meta-learning MAML: https://arxiv.org/pdf/1703.03400.pdf Reptile: https://d4mucfpksywv.cloudfront.net/research-covers/reptile/reptile_update.pdf 1) do I ...
2
votes
0answers
9 views

Is there a formal relation between weight regularization and compression?

In my understanding, compression, strictly speaking, means that we diminish the amount of data required to describe something, such as a model. E.g. compressing an image file means to create a file ...
3
votes
0answers
52 views

Does anybody know this measure of model fit / prediction error?

Let $y_i$ be the true value and $\hat{y}_i$ a prediction from a model. Then, for example $$B=n^{-1}\sum_{i=1}^n \hat{y}_i - y_i$$ is the prediction bias and $$MSE=n^{-1}\sum_{i=1}^n (\hat{y}_i - y_i)^...
3
votes
1answer
40 views

Topological rather than metric based machine learning theory?

The first notion of continuity in a math class is usually the one based on metric spaces. In particular, the $\epsilon,\delta$ definition of continuity. But in topology, a more general notion of ...
1
vote
2answers
95 views

Time Series Analysis for a Newbie

I am a beginner in time series analysis and machine learning. I have a dataset where I want to analyse and predict a time series data. I have a pollutant variable and four meteorological parameters ...
0
votes
0answers
16 views

Selecting the best subject's data and features to optimize the analysis

I am not good at statistical analysis. So I am posting here my case and looking for your kind suggestions. My case: I have data from subjects, which each subject has two similar runs that were ...
1
vote
0answers
37 views

Which is faster: a bank with five lines of ten or one line of fifty?

I'm working on a probability question with mean and variance. Let's say that I have two banks. They are identical in every way, except that bank A has five lines with ten people each and bank B ...
6
votes
0answers
176 views

Recreating figure from Elements of Statistical Learning [closed]

I am trying to recreate FIGURE 3.6 from Elements of Statistical Learning. The only information about the figure is included in the caption. To recreate the forward stepwise line my process is as ...
1
vote
1answer
30 views

Creating Target Variable for time series change point detection

I am working on a time series data for which I intend to impliment machine learning model for detecting change point in time series data. This data is recorded fom machinary and we have to predict ...
0
votes
0answers
5 views

Why 'sequentialfs' of MATLAB stops before the optimum feature subset is selected?

I am using "sequentialfs" of MATLAB to select features from 271 features for 871 subjects over 2 classes. I used the backward sequential function. I noticed the features selected after using the ...
0
votes
0answers
18 views

Implications of deploying a predictive model overfitting training data but consistent in validation folds (classification)

If a model is build on very dirty data, it is common to not be able to prevent an overfitted result even with rigorous regularization attempts. However, it is also common that some lift-producing ...
0
votes
0answers
29 views

When to use which classification model?

This is something that continues to give me trouble. Assuming I am working to extract a classification from a dataset and assuming I have the computing resources to do the necessary calculations (in ...
0
votes
0answers
8 views

Determine confidence in estimate from past estimates

I have a population on the order of 10,000 samples. These samples represent individual estimates of a variable, and for each estimate I also have the actual value of the variable. For each sample I ...
0
votes
0answers
18 views

About the time differences in the Bellman equation

I am trying to grasp fundamental mathematics behind the Reinforcement Learning and so far I have unterstood how the Value Iteration and Policy algorithms do converge (contractions, etc.) I have still ...
5
votes
1answer
43 views

What is meant by 'Black box variational inference'?

I'm aware of the topic of variational inference (VI) however I'm not really sure what Black box VI is? In particular I am watching a video by David Blei titled Black box variational inference and on ...
0
votes
0answers
13 views

are there any supervised learning methods that can only be applied to data with continuous features?

Are there any methods that exclusively work on continuous features? At first i imagined that linear models would demand this, but discrete values can be transformed and encoded such that they can be ...
1
vote
1answer
32 views

How are the weights updated in the perceptron learning rule?

I'm considering a perceptron model. I know that when feeding observations from the training dataset to the model, if the model correctly classifies the input, then the weights for this input will not ...
0
votes
0answers
24 views

Dependent and Independent Features

I have a data set that has 3 columns(Features), I would consider every feature to be independent. Col1, Col2, Col3 would all be independent, but Col4 would be the features strung together which would ...
3
votes
1answer
72 views

Gradient Descent and Back Propagation in Neural Networks: derivation in vectorised form

I was learning deep learning from Andrew Ng's course on Coursera and in one of the programming assignments to code out Neural Networks from scratch, the formula for the derivative of cost function wrt ...
0
votes
0answers
30 views

What are the best classification models with one feature?

I am looking for some insight on making a model with just one predictor. Let's assume the data are not linearly separable (because otherwise I assume it wouldn't matter which linear method to use). ...
0
votes
0answers
33 views

For p>>n, where p is number of predictors and n is number of observations, is n P or P+N?

For p$\gg$n, where p is number of predictors and n is number of observations, is n P or P+N? For example, if I am building a binary classifier, and I have 165 positive and 165 negative observations, ...
3
votes
1answer
58 views

Find true negatives in a confusion matrix

I'm trying to find the True negative in a confusion matrix, I have computed successfully from scratch the precision and recall/sensibility, now i need to compute the accuracy and specificity. This is ...
1
vote
0answers
13 views

How are the various guarantees provided to SVMs by Statistical Learning Theory affected by the Kernel Function

I never studied the field in depth, but I am very aware that state of the art performance in most ML tasks is now achieved by various flavors of neural networks. At the same time, Vladimir Vapnik, ...
1
vote
1answer
25 views

Bypassing inverse matrix calculation and the comparison of Gradient Descent and Newton Descent

I am currently reading Pattern Classification (2nd ed) by R. O. Duda, P. E. Hart and D. G. Stork, John. In the 5th chapter the Gradient Descent algorithm is introduced with the following notations : $...
0
votes
1answer
16 views

How to deal with different image sizes during training and inference? (e.g. Stacked Hourglass)

In some of computer vision papers I read that they start off with a bigger sized image and use pooling to reduce dimensionality and train on the image with lower resolution. However, they don't ...
1
vote
0answers
21 views

multivariate multiple regression with probabilities as the dependant variable

I have a problem that involves multiple dependant variables, where each dependant variable is a probability and for each observation the probabilities sum to 1.0. I also have a range of independent ...
1
vote
1answer
20 views

Why not sample action from Q values?

When collecting experience from which to estimate a Q(s,a) function, one common technique in the literature is to follow an epsilon greedy-strategy. In this strategy, the agent selects a random action ...
0
votes
0answers
10 views

Training Hidden Markov model with GMM, nan appears after some iterations

Problem During the training process of my continuous observation sequence data using HMM with GMM mixtures, the cost function reduces gradually and it becomes NaN ...
1
vote
0answers
32 views

Gradient Descent Vectorization with Numpy (1D transpose confusion) [closed]

I'm working through Andrew Ng's original Stanford course and ran into some numpy confusion. Basically, my main question is, if we dot product a 1D array with a 2D array in numpy (and the dimensions ...
2
votes
1answer
20 views

An artist published M albums. How to make a selection from the album so that all my N favorite songs are covered, while minimizing my cost?

This is a real life question.. I have a list of N favorite songs from an artist. Out of all M albums from the artist ever published,I want to buy a few albums to cover all of my N favorite songs, but ...
0
votes
1answer
9 views

Why disentangled factors learnt by unsupervised models could meet human criterion of high-level or semantic meaning?

Multiple generative models (beta-VAE, InfoGAN, Glow, etc.) claim the capabilities to disentangle and control high-level factors in their generated samples. For instance, for a VAE decoder network F(z) ...
2
votes
2answers
43 views

Why in CNN network 3 filters are learning same feature?

I am using a simple CNN with one convolutional layer and one fully connected layer. I am using 3 filter channel and one input channel. I run my code 500 times with random initialization of weights in ...
3
votes
1answer
31 views

Linear Regression Assumptions vs. Gauss Markov Theorem

I am wondering what is the difference between the Gauss Markov theorem and the assumptions of linear regression found here or here? For example, the third link says that the distribution of residuals ...
1
vote
0answers
25 views

Equivalence of variable selection criteria in forward stepwise regression

Say we have already selected $x_1$ through $x_k$. To select the next variable, forward stepwise regression either: a. picks the variable that when added to the already selected variables, gives ...
2
votes
1answer
46 views

Poisson regression predictions

I've started looking at GLMs and I've worked out point estimates for Poisson regression using the canonical exponential link function. So the likelihood being $$ \ell(y_i \vert x_i, \beta) = \sum_{i=...
0
votes
0answers
8 views

Quantify the over/underrepresentation of data satisfying some condition(s)

I am doing binary classification on a dataset. It has the following distribution across a certain feature that takes on 13 possible values: Then I have this distribution which is a subset of the data ...
0
votes
0answers
20 views

What is wrong with tuning parameters on training set as opposed to validation set?

When creating a machine learning model it is suggested to split your data into train, validation, and test sets. Here is my understanding of what they are for. Train: Use this to train the different ...
1
vote
0answers
11 views

Use standard scaler in combination with min max?

I asked a colleague if it would be wrong to use feature scaling in decision trees since it is not required by the algorithm. He said that not only I could, but also it is a good practice. However, he ...
1
vote
1answer
41 views

What machine learning/deep learning technique to use for a data set consisting of 2D plane [closed]

I'm new to machine learning/deep learning field. In an assignment I've been given a data set of 1000 data points. Each data point is a 2D place of dimension 1024 x 1024. In each plane there are some ...