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
4 views

What are the rules for evolving a partially connected neural network using a genetic algorithm?

I am working on a project where I need to implement the NEAT algorithm in python. after doing some research I came across an issue that I can't seem to find a solution for online, I hope this is the ...
1
vote
0answers
5 views

Defining an agent's action as a parameter approaches a specific value

Python newbie here, please bear with me. I'm trying to code a simple reinforcement learning program in which an agent repeatedly chooses from two different gambles of the form: win $x_1$ with ...
0
votes
0answers
7 views

Loss low but extremely low feature importance score

I'm running permutation importance on DNN and for some reason the numbers seem suspiciously low, highest scores are around 0.015 with explained variability score and pretty much the same with r2 ...
0
votes
0answers
5 views

Compare text corpora

I have am currently performing speech recognition experiments on 2 different corpora. I have the ground truth human-labelled texts for both corpora. I am performing different experiments that lead to ...
2
votes
1answer
16 views

Repeated Nested Cross validation

I'm aware that nested cross-validation is used for hyperparameter tuning and model selection and that repeated k-fold cross-validation is used to improve the estimated performance of the model. My ...
0
votes
0answers
17 views

Finding the leave one out cross validation error rate based on the graph

I have a K-nearest neighbor classifier that was trained using the data, except for that point and then used it to predict the label for the withheld data point. I need to find the leave-one-out cross ...
0
votes
0answers
7 views

How to combine multiple implicit feedback

If there are multiple different types of implicit feedback signals, what would be a good way of combining them? For example, suppose that you have "click" and "thumbs up/down" ...
0
votes
1answer
12 views

How can i cite Sequential Forward Feature Selection (SFFS)?

I've seen many papers/books about this technique but none cite its author. Is it ok to cite any machine learning theory book that explains it? Thanks.
1
vote
0answers
6 views

Transfer Learning: data in the source domain and the target domain are required to be independent and identically distributed

In instance-based transfer learning, it is said that data in the source domain and the target domain are required to be independent and identically distributed. When it says that the data "are ...
0
votes
0answers
24 views

Why does my neural network only approximate my function for the first half of my data?

I recently tested out scikitlearn and trained an MLPRegressor. I didn't do anything fancy, creating a basic NN with a few layers and a good amount of neurons, and then tested it on a very simple ...
0
votes
1answer
28 views

How to check to use linear models for huge data?

Generally, whenever we had the data when we want to use linear models we try to make numerical features to have normal distributions. And we check it with by plotting distribution or qqplot. It's the ...
0
votes
1answer
31 views

Counterintuitive coefficients in elastic net logistic regression

In a model run of elastic net logistic regression, I encountered a very counterintuitive coefficient. I have looked into the data, model and script, but, I still cannot seem to wrap my head around the ...
0
votes
0answers
17 views

Regression analysis of data while there is no dependent variable

I am a master student and I started my training in a laboratory to work on Ecology. We are considering some polluting gases(=4) in air and we measure their ...
0
votes
0answers
10 views

convert classification model into object detection model [closed]

I have created a Custom CNN for classification using ImageNet like custom Dataset(39 x 39px images) with Keras-Tf2.2.0. I have got 98% classification accuracy. how Can I use the same classifier for ...
0
votes
1answer
36 views

How can you use a neural network to extract the needed information from social media ads?

How to solve the following problem using neural networks / and machine learning / artificial intelligence? Input data - is an ad from a channel or group of a social network. For example, this: A room ...
0
votes
0answers
34 views

How do I use multiple rows in a frame to get one classification?

For example, I have ~500 5x5 data frames like the following: ...
0
votes
0answers
11 views

Deriving vectorized form of linear regression

We first have the weights of a D dimensional vector $w$ and a D dimensional predictor vector $x$, which are all indexed by $j$. There are $N$ observations, all D dimensional. $t$ is our targets, i.e, ...
1
vote
0answers
13 views

Difference between One Rule Classifier and Decision Stump in WEKA

WEKA Explorer seems to come up with two different models for OneR (rules) and Decision stump (trees). Is has to be the underlying measure of "best split" that is different. But for a single ...
0
votes
0answers
6 views

how to do the feature_selection in testing data

I have a simple RF classifier model trained with a sample dataset and it works fine. So, I use some test data to predict the target class and let's say it find the target class as ...
0
votes
0answers
9 views

Leverage Statistic (h) for Multiple Predictors [duplicate]

I'm working through Introduction to Statistical Learning and understand the leverage statistic for a simple linear regression. However the text says "There is a simple extension of h_i to the ...
2
votes
1answer
23 views

How is prior distribution of weights selected in MAP estimates?

I just read MAP estimate of linear regression , and got to know that the regularization term is the result of considering prior distribution of weights . So , my question is how is this prior ...
0
votes
0answers
10 views

What is the wide scaling mentioned in Efficientnet paper?

I do not quite understand what's the specific method of wide scaling? multiscale of the kernel or small stride?
0
votes
0answers
8 views

Is there a UCB type algorithm for linear stochastic bandit with lasso regression?

Why is there no upper confidence bound algorithm for linear stochastic bandits that uses lasso regression in the case that the regression parameters are sparse in the features? In particular, I don't ...
1
vote
0answers
46 views

An interesting non-smooth regression

Consider the following setup: Let $x_1^k$ and $x_2^k$ be length $N$ vectors of observed reals, where $N$ is about, say, 100,000, $k\in 1:K$, and $K$ is about, say, 200. (So $x_1,\;x_2$ can be thought ...
4
votes
1answer
32 views

How to use an F-test to understand if a categorical input is statistically significant?

I am trying to understand F-tests in general and how to apply one for a particular problem for a data set in "Introduction to Statistical Learning with Applications in R". The data set is ...
0
votes
0answers
6 views

Should I use disjoint time periods to predict loan default?

I'm relatively new to machine learning (so pardon me if I'm being naive). I am trying to build a model to predict loan default using only the information available at origination. I figure my model ...
0
votes
0answers
16 views

Is my understanding and presentation of concept of Gradient Boosting correct?

Initially the model is trained with a training set $\{x_{i}, y_{i}\}_{i=1}^{n}$ by minimizing a differentiable loss function $L(y, F(x))$, and, is initialized with a constant value, \begin{align*} F_{...
0
votes
0answers
8 views

Can the label powerset method be used to predict probability for reach class?

The label powerset is a method used to transform a multi-label problem to multi-class problem. The idea is straightfoward, just enumerate all the possible combinations of outputs, and treat each of ...
0
votes
0answers
12 views

Can $f$-divergences narrow the discrepancy between train and test fits in machine learning?

Machine learning models whose task is to predict unseen test data would work best if the test data's distribution turns out to be the same as the training data's distribution. Real data seldom works ...
0
votes
0answers
11 views

Best strategy to run thousands of small tests on a machine with one GPU and multi-core CPU [closed]

I have recently got a new machine with GPU and multi-core CPU. I have thousands of small tests (one minute each) to run in order to tune my model and extract some other information. What is the best ...
2
votes
0answers
15 views

ANOVA vs. mixed models

I'm confused between the differences between x-way/mixed ANOVA models and mixed models. Is there a difference? If so, what is the difference and why?
0
votes
0answers
22 views

Expected Risk vs True Risk [closed]

I wanted to know if the true risk is equal to expected risk? In our script we define the expected risk: $$R(h):= \mathbb{P}[h(x)\neq y] = \mathbb{E}[I[h(x) \neq y]]$$ The true risk is according to ...
0
votes
0answers
11 views

Should convolutions or transposed convolutions be used in the decoder part of a Conv-based autoencoder?

I am implementing a convolutional autoencoder. For the decoder part of the model, some examples (such as this one from Francois Chollet) use standard convolutional layers (Conv2D in keras) in the ...
1
vote
0answers
16 views

Should we train base learners on same folds when we stack different models

I am confused about k-fold stacking. Should we train all the base learners on same folds? I mean is it ok to do the KFold split with different seeds or all base learners should trained on folds ...
0
votes
0answers
17 views

How to prepare data for Bert fine-tuning?

I am looking at Bert documentation in order to fine tune a pretrained model... So lets say I have a new dataset for paragraph classification. I get a vector for each paragraph and I use a simple KNN ...
0
votes
0answers
20 views

Loss function for regression problems to penalize negative predictions?

I am working on a regression task where the target variable cannot be strictly negative. To do the predictions, I am using the LGBM framework (Python) with the RMSE loss. Issue that I am facing is ...
1
vote
2answers
26 views

Dealing dataset with NAs (MNAR)

I have a dataframe with many (>50%) NAs values and I am looking for a way to deal with it. From what I've found, I think many people recommend using imputation like multiple imputation or using ...
1
vote
0answers
11 views

Search depth of Alpha Go and Alpha Go Zero [closed]

I cannot find reliable sources but someone says it is 40 moves and someone says it is 50+ moves. I read their papers and they use value function (NN) and policy function to trim the tree so more ...
0
votes
0answers
6 views

how to understand VAE latent distribution?

My question is regarding variational autoencoder latent space distribution. I understand that with KL loss added in the loss function, after adding KL loss to the loss function, p(z|x) will be forced ...
1
vote
1answer
42 views

Understanding mini-batch gradient descent

I would like to understand the steps of mini-batch gradient descent for training a neural network. My train data $(X,y)$ has dimension $(k \times n)$ and $(1 \times n)$, where $k$ is the number of the ...
0
votes
1answer
22 views

Understanding the Q-learning loss function?

Perhaps this can be explained a little more to me. I understand what's in literature but I'm struggling to understand why this is the preferred loss. If we have an agent that can move ↑↓→← and for ...
0
votes
0answers
12 views

Image likelihood associated with cross correlation loss in image registration?

In this question I'll represent the fixed image as $\bf{x}$, the image we are trying to align onto it as $\bf{y}$, and the transform warping $\bf{y}$ onto $\bf{x}$ as $\phi_z$, parameterized by $z$. ...
3
votes
1answer
27 views

Modeling and counteracting exposure bias in recommender systems

I am looking for best strategies to train a new recommendation model from the biased data (due to modeling bias from the previous model). For e.g. Lets assume I have an e-commerce site and initially I ...
0
votes
0answers
12 views

Binary probability scoring: Intuition on why a method might perform better in terms of Brier, log loss but worse in terms of Area under ROC/PR curve?

I'm trying to compare two methods. I have surface knowledge about these scorers, so I've noticed that scorers in which method A performs better are both proper scoring rule, while B performs better in ...
0
votes
1answer
30 views

Time Series Forecasting: ARIMA\VARIMA vs Machine Learning \ Deep Learning

I am working on the development of a time series forecasting, and I have some doubts on the model I should use to achieve better results. PREMISE: Multivariate Time Series: my time series is a ...
0
votes
0answers
5 views

Dirichlet process Posterior notation

While reading through Yee Whye Teh's tutorial on Dirichlet Process, I came across the Posterior distribution for DP. After observing some draws $\theta_1,\theta_2, ... ,$, the posterior is defined as $...
0
votes
0answers
12 views

Multiclass classification into elements of the Cartesian product of classes

I want to classify samples in m x n x p classes, which are the tuples that belong to the Cartesian product of three sets A, B, C with ...
0
votes
0answers
10 views

Aggregation model in stacking with or without initial features

I am implementing a Stacking Ensemble, which works in general, for Supervised problems. Ideally, this was my idea: Train: I take the training-set and after training the N models of the ensemble, I ...
0
votes
0answers
7 views

Measure of base distribution $G_0(A)$ in the Dirichlet process

I am studying the Dirichlet Process and I came across this rule that states that given a partition $A_i's$ over the space $\Theta$, which represents the space in which $G_0$ is drawn from, $$(G(A_1),G(...
1
vote
1answer
17 views

Minimum value of cross entropy cost function

If I understand right, general cross-entropy cost function can be written as: $$c := - \sum_{i} t_{i} \log (a_i)$$ where vector $\mathbf{t}$ is 'true' discrete pdf and the vector $\mathbf{a}$ is the ...

1
2 3 4 5
314