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.

Filter by
Sorted by
Tagged with
0
votes
0answers
14 views

What is the best way to avoid/deal with tie-breaking when using majority voting ensembles?

What is the best way to avoid/deal with tie-breaking when using majority voting in ensemble? I tried to use Weighted majority voting; however, the F1 score dropped significantly. Are there any other ...
0
votes
0answers
19 views

Linear regression classifier [closed]

A linear regression classifier is designed by selecting 5 features of 10 possible, using a set of 50 observations. Which is the best method for selecting features?
0
votes
0answers
13 views

Pseudo log-likelihood computation in Restricted Boltzmann machines

I am writing a RBM and I have a doubt. Searching on the Internet, I found that the pseudo log-likelihood is often computed during training. The code I found is something like the following one. ...
0
votes
0answers
12 views

Is there any term for tailoring data to the suitability of different models?

Is there any literature or fields I could start with, if I was asked grade each of a bunch of datasets, or assess their suitability for different model types, and inference goals?
0
votes
0answers
12 views

Why is ancestral sampling used in autoregressive models?

I have been reading about autoregressive models. Based on what I've read, it seems to me that all autoregressive models use ancestral sampling. For instance, this paper says the following in Abstract: ...
1
vote
0answers
8 views

When Should we apply the Aggregations(mean, sum , count, max, min) and how to deal with correlations features

I am beginners to machine learning , I worked on the some basics machine learning problems. I had a little bit confusion about feature Engineering. some people are using aggregations functions for ...
0
votes
0answers
11 views

What is the difference between spliting the dataset into training and testing or collecting the training and testing data seperately?

I am working on active learning and I was wondering about the difference if we split the dataset into training and testing or collecting and labeling the training and testing datasets separately. ...
0
votes
0answers
16 views

How to deal with the problem of correlated samples from time series in machine learning

We know that most of the machine learning algorithms assume the samples are i.i.d. However if the samples are from time series (like ARMA in the prediction of time series), then they are highly ...
0
votes
0answers
7 views

Independent component analysis when the distribution of components is known?

I was wondering how can we conduct ICA when the distribution of all components is known. Usually, people will assume that this problem is much easier than the original "blind" source ...
0
votes
0answers
9 views

Boltzmann Machine weight update scheme [closed]

I would like to implement Boltzmann machine with two hidden and two visible units. The four possible hidden units configuration are $(0,0), (0,1), (1,0), (1,1)$, and their probability distribution is $...
5
votes
1answer
94 views

Can proportions over variables be learned/predicted with Neural Networks using multiple-outputs?

I am interested in understanding how Neural Networks could be used to both learn from and predict proportions. That is, say matrix $X$ is the training features data with $N$ cases and $k$ features. ...
0
votes
1answer
13 views

How to use data augmentation in the context of model evaluation in machine learning?

I'm trying to use data augmentation for training a model for a classification task. But I'm not sure about how to use data augmentation in a fair and meaningful way in the evaluation of a machine ...
0
votes
0answers
29 views

Does Random Forest model require similar sample size across different sample?

Suppose I have a dataset consisting different fruits: 60 apples, 100 oranges, 120 bananas, 7 grapes, 900 pears, I want to train a random forest model using these fruits, but what should i do with ...
7
votes
1answer
525 views

Modern machine learning and the bias-variance trade-off

I stumbled upon the following paper Reconciling modern machine learning practice and the bias-variance trade-off and do not completely understand how they justify the double descent risk curve (see ...
0
votes
0answers
14 views

How topological fingerprints are effectively used in a Machine Learning model [closed]

I was just perplexed about the practical usage of topological fingerprints coming out from persistence homology approaches. Once I've obtained persistence diagrams, how do I effectively use them to ...
1
vote
1answer
12 views

Evaluation metric for time-series anomaly detection

My dataset is time-series sensor data and anomaly ratio is between 5% and 6% 1. For time-series anomaly detection evaluation, which one is better, precision/recall/F1 or ROC-AUC ? When empirically ...
2
votes
0answers
10 views

What is the “problem of adaptive estimation” that necessitate the development of honest tree?

My question is about the Athey & Imbens (2016) paper. Even though this paper develops honest tree to estimate heterogeneous treatment effect (HTE), in this question I'm only asking about the ...
-1
votes
0answers
16 views

In LSTM, is hidden state of each neuron sent to other neurons in the same layer?

In LSTM layer, is each hidden state fed back as input for each neuron separately or for all neurons of the same layer? In other words, which figure is correct: Figure 1 or Figure 2?
0
votes
0answers
17 views

nested CV for parameter importance

I was wondering whether you can help me sort out some confusion I have about using nested cross validation (i.e. inner + outer cv loops) in a ML analysis. I am happy to share some code, but I think ...
1
vote
0answers
12 views

Discrepancy in marginal likelihood in non-stationary gaussian process

I am working on implementing a variation of the Gibbs kernel (Eq. 4.32 in GPML book) proposed by Plagemann et al.. The authors propose a method to learn latent length scales at $m$ locations ($m <&...
0
votes
0answers
7 views

Practical calculation of EER (equal error rate) for biometric tasks

I am recently experimenting with the speaker recognition task. So, EER is calculated for a threshold FAR = FRR. Now, my question is how can I calculate this given I ...
4
votes
3answers
445 views

Autoencoders as dimensionality reduction tools..?

I'm trying to get a basic understanding of AutoEncoders. Basically they are neural networks with a very low representation of the original input in some hidden layer, and then a final layer which has ...
1
vote
0answers
21 views

What does it mean if the distribuctions of TP and FP are really similar?

I was training a Neural Network these days, and I plotted the values of TP, FP, TN, FN on the validation set as the training progresses. I found that on the validation set, the distribuctions of TP ...
0
votes
0answers
8 views

Using compositional data analysis to represent chemical compounds

I've recently got some insights about compositional data analysis, wondering whether it could be suitable for the framework I'm currently in. Recently, I've been very interested in trying to find some ...
3
votes
1answer
32 views

MOOC on Causal analysis - Recommendations for software engineer

Am a software guy with no background in causal inference. While I am now familiar with prediction techniques due to plethora of courses available online, I would like to seek recommendations from ...
0
votes
0answers
19 views

Sort items based on frequency and recency [duplicate]

I'm working on a problem which requires me to sort a list of static items for each user. I understand best way to solve this problem would be to come up with a function that captures both the ...
0
votes
3answers
32 views

Is this Keras model overfitted?

I am currently writing my thesis on deep learning models where I train a VGG like model. I trained my model always with Early Stopping function from Keras, where it stopped training after ...
0
votes
0answers
7 views

Univariate multi-step prediction model for missing value imputation

I have a dataframe with columns of timestamp and energy usage. The timestamp is taken for every min of the day i.e., a total of 1440 readings for each day. I have few missing values in the data frame. ...
2
votes
1answer
23 views

Cost of Nested Cross-Validation

What is the cost of nested cross-validation in terms of the number of times the algorithm needs to perform a fit-evaluate step? Based on this description of the algorithm, I think the answer is: $n \...
5
votes
2answers
48 views

Is there something like a confusion matrix for a probabilistic score rather than categories?

Imagine we have pictures of three animals: dogs, cats, and horses. We train our image classifier and get a confusion matrix, noticing that the model tends to predict that dogs are horses. But then we ...
1
vote
0answers
11 views

Methods for Predicting Destination Locations based on Start Locations in 2D Space

I have location data $(x_1, x_2)$ and the associated labels $(y_1, y_2)$, which can be interpreted as origin and destination points in physical space. Now I would like to predict destination locations ...
0
votes
0answers
7 views

Problem with scikit-learn! ValueError: Input contains NaN, infinity or a value too large for dtype('float64') [closed]

https://github.com/raimondo82/intentions_survey/blob/main/Intentions_survey.ipynb Hello, I am having this problem with scikit-learn when trying to get recursive features elimination with cross ...
0
votes
0answers
21 views

Information Value vs Variable Importance

I'd like to understand the differences between these two concepts: Information Value vs Variable Importance. I understand that they can be used for feature engineering purposes but what is the main ...
1
vote
1answer
16 views

Categorical data treatment courses

Is there a book or online course/mooc related to the treatment of categorical data? I have seen this course as an elective in my university in the Statistics degree, and also as a core course in a ...
1
vote
2answers
27 views

Can a neural network nodes “underweight” or “overweight” themselves?

I was under an impression that artificial intelligence is modelled after organic intelligence. Under the context of organic intelligence, it seems that some individuals are capable of getting caught ...
0
votes
0answers
25 views

Is it possible to have one general accuracy for a number of class [closed]

I have obtained confidence score (softmax probabilities) of 100 different classes for 20 different convolutional neural network models. After obtaining confidence score of 100 classes from different ...
0
votes
0answers
15 views

Would a machine learning algorithm run faster on a pre-clustered dataset than its non-clustered equivalent? [closed]

Suppose we want to perform regression/classification on some arbitrary dataset. I read somewhere (can't recall where) that clustering is sometimes used as a preprocessing step before further learning ...
0
votes
1answer
32 views

Why is there a ZeroR and OneR classifier but no TwoR classifier?

I am using WEKA, and I noticed that there is a ZeroR classifier and a OneR classifier. The ZeroR classifier always predicts the majority class, while the OneR classifier bases its predictions on only ...
0
votes
0answers
10 views

EM algorithm common errors

I have been trying to use EM algorithm to estimate a modification of normal mixtures with means (and variances) depending on a set of observables. My code works reasonable well against simulated data, ...
4
votes
3answers
346 views

Why isn't the ROC curve naturally plotted in 3D? [duplicate]

Something that really confuses me with how ROC plots are generated is that, according to Wikipedia: The ROC curve is created by plotting the true positive rate (TPR) against the false positive rate (...
4
votes
1answer
53 views

What is the purpose of conducting Simple Random Sampling WITH Replacement?

Part of data preparation is simple random sampling. Random sampling can be of two forms with replacement or without replacement. With replacement, subset sampling simply might contain duplicates of ...
0
votes
0answers
8 views

how to train the last n layers in VGG16 model? [closed]

I'm somewhat a beginner here. I'm making an image classification model for 3 classes using transfer learning and I wanted to train the last 4 layers of vgg16 without any additional layers. but since ...
0
votes
0answers
24 views

Main idea behind reparametrization trick (distribution to function)

If I got the idea correctly, one of the main concepts behind the reparametrization trick, first presented in Kingma, D. P., & Welling, M. (2013), Auto-encoding variational bayes (ArXiv Preprint ...
4
votes
0answers
27 views

Re-sample train data so it represents real-world data

I have a classification problem - will a student pass a course, or not. I have real-world data consisting of million of students enrolling in my course. In addition, I have 1,000 tagged students - ...
1
vote
0answers
17 views

How autocorrelation work based on the data plot?

Let's suppose we have a time series is a=[1,1,1,3,3,3,1,1,1,3,3,3] as then the autocorrelation figure for this time series is The lag here is 4, for lag = 1, the autocorrelation is between [1,1,1,3,...
0
votes
0answers
16 views

batch renormalization questions

I was going in details through paper about batch renormalization (arxiv link). I don't quite understand two things there. Maybe there is anyone who faced similar issues / knows the answer and could ...
0
votes
0answers
10 views

Issue with “Length Mismatch Error” [closed]

I was trying this code and manage to get expected resul as seen in the picture below the code ...
1
vote
0answers
27 views

Are t-SNE and UMAP for dimensionality reduction compatible with cross-validation in machine learning applications?

t-SNE: Based on how I understand the original t-SNE algorithm, it requires a whole dataset for doing the transformation. That is, there are no distinct "fitting" and "transformation&...
0
votes
0answers
3 views

Mask for image padding in semantic segmentation

I'm using data augmentation for a semantic segmentation task, where some images are cropped or rotated. As a result, some padding is added to ensure that the image is always the same size. These ...
0
votes
0answers
5 views

sample complexity for class of conjunction of atmost n boolean literals

Why is the no. of functions or $|H|$ taken as $3^n$ while calculating the sample complexity bound and why not $3^{3^n}$ based on the general formula of $k^{k^n}$ where n is the no. of variables in a k-...

1
2 3 4 5
343