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
1answer
23 views

How come my accuracy is so oscillatory?

I'm training up a linear regression model on some vehicle data (sorry I can't be more specific than that). I'm experiencing some very strange output on the accuracy of the model and I'm not sure it's ...
2
votes
1answer
17 views

How to draw ROC curve for a multi-class dataset?

I have a multi-class confusion matrix as below and would like to draw its associated ROC curve for one of its classes (e.g. class 1). I know the "one-VS-all others" theory should be used in ...
0
votes
0answers
6 views

Any benchmark dataset for multi target regression problem?

If the question does not go with the Cross Validated, please, kindly remove the question. As I have found people got response about dataset related question in here, that is why I am giving a post. I ...
0
votes
1answer
17 views

Textbook recommendations covering machine learning techniques for causal inference?

Over the past 15 years there has been progress in adapting machine learning methods for causal inference. For example: targeted learning, double machine learning, causal trees. Is there a textbook ...
1
vote
0answers
26 views

Bias vs. variance

I have a question about bias/variance trade-off for different competing models. Say one has estimated model A and model B and calculated their respective train and test error. How does one yield an ...
1
vote
0answers
15 views

Finding the unknown correlations to data from an image

I have a complex image of a biological device, and I have the results of the device being used as a data frame collected from other sources. I want a predictive model to say this image will create ...
0
votes
0answers
18 views

Are there systematic approach / tests to interpret the performance of a Machine Learning model on particular training and testing dataset? [closed]

I was wondering if there are any known procedures one can follow to systematically interpret / explain the performance of a machine learning model with the training and testing data? For example, what ...
0
votes
1answer
11 views

How to approach this: Percentage change in one KPI leading to change in other KPIs?

I want to know how can I approach or model this problem. I have 7 KPIs (3 of them dependent on each other) and one main KPI (total 8 KPIs). I want to understand effect of these 7 KPIs on the main KPIs....
0
votes
0answers
14 views

Time Series: Do I understand Windows Slicing correctly?

So in the following Thread it is discussed about augmentation for time series: Data Augmentation strategies for Time Series Forecasting The first answer refers among others to the following: Window ...
0
votes
0answers
8 views

From nn.MaxPool3d to "strided convolution layer"

I am working on GAN for medical images and the generator mode is Unet, but some bugs crash me. like Unet's task can be adversarial learning and semantic segmentation. but it seems some layers are ...
1
vote
0answers
13 views

How to encode variable-length unordered data

I am trying to learn an encoding of data, but I want to do so in a way that doesn't depend on the order of one of the inputs. The data is given by $\{(X_i,y_i)\}_{i=1}^{n}$ where $n$ is the number of ...
0
votes
0answers
27 views

Question related to BFGS algorithm in "optimx" package [closed]

I would like to print out all intermediates estimates for the parameters in a logistic regression using the "optim" function from the "optimx" R package. How can do it ? For ...
0
votes
1answer
30 views

Monte Carlo simulation on machine learning classification [closed]

I have done training different kinds of machine learning classifiers (e.g. logistic regression, SVM, random forest, etc.) and the data used is heart failure comprising of 13 columns and 299 rows. The ...
0
votes
0answers
18 views

Machine Learning with similarities as inputs

Assume I have wide data e.g. 1.000 examples 10.000 features. I want to train a machine learning model, for example, a neural network. Instead of learning in the feature space, can I learn in the space ...
0
votes
0answers
12 views

Improve model accuracy in multi-classification problem

I use a MLP to classify three different classes A, B, C. The loss function I use is categorical cross entropy and the optimiser ...
1
vote
0answers
34 views

Is there a good replacement of sum of squared deviations that do not tend to split on edges?

I build a predictive model (regression) on a dataset that has just one real-valued feature and one real-valued target. To make it even simpler I want to find just a step function (decision tree with ...
0
votes
0answers
12 views

ARIMA forecast and detect out-of-sample outliers

I have a question about the outliers detect in out-of-sample data (or imagine the data add new value per minutes like stock price or something else). First I using in-sample data to build the model (...
0
votes
0answers
11 views

What is the term for an algorithm that satisifies the bound $\lvert l(z,h_{s})-l(z,h_{s^{'}})\rvert \leq \beta$ where samples differ in one component

Consider $\mathcal{A}$ as an algorithm that satisfies the following condition for the loss function where $l$ represents some loss function and $z\in \mathcal{X}\times \mathcal{Y}$ is a sample and $s$ ...
0
votes
0answers
18 views

Search using raw word embedding similarity from BERT [closed]

I have a list of about 100k foods I want users to be able to search through. I’ve explored using word2vec to map search terms and food names to vectors, then return results via vector similarity (...
0
votes
0answers
9 views

Google Activity Recognition/Transition API disproportinate ENTER and EXIT events

I am working on a project to detect user depression from smartphone sensor data. One such sensor data is Android Activity Transition events such as WALKING_ENTER, WALKING_EXIT, RUNNING_ENTER. I ...
0
votes
0answers
3 views

Do Self-Attention GAN models belong to Autoregressive models group?

Do Self-Attention GAN (SAGAN) models belong to Autoregressive models group?
1
vote
0answers
20 views

Principal Components Regression

We want to perform regression using PCR (Principal Components Regression) and we have two of our variables as X1 and X2. The correlation between X1 and the outcome variable Y is 0 and the correlation ...
0
votes
1answer
15 views

Uncertainity band in Neural networks

I am working on a problem where I have to give the uncertainty band of my predictions like the image attached. I have seen a StackExchange solution for this, but in the solution code, we are using ...
1
vote
1answer
18 views

Sign change in LASSO and RIDGE of coefficients

I am estimating in total three models: Logistic regression without any penalization (as benchmark model), logistic regression with L1 penalization (LASSO) and with L2 penalization (RIDGE). Now i ...
0
votes
1answer
119 views

How to add and vary Gaussian noise to input data

I have a time-series data and I would like to add an additive Gaussian Noise to the input of the data. What I am trying to do is that I want to test my ML predictive model against different level of ...
2
votes
1answer
41 views

Distributional Assumptions in Machine Learning

In more classical statistical methods like linear regression, we can quantify how well our model generalizes under certain strong assumptions. For example, we know that $\hat Y = X \hat \beta \sim \...
1
vote
1answer
18 views

non-time related features not contributing in ML forecasting

I made some ML models for a demand forecasting problem. My predictors were the time-related features derived from the DateTime column (as in 'week of the year', 'day of the month', 'day of the week', '...
0
votes
0answers
43 views

Applications of the KL-Divergence in Real World Predictive Models [closed]

This is a question that I have always had. Take the KL (Kullback-Leibler) Divergence - the official definition of the KL Divergence (according to wikipedia): "In mathematical statistics, the ...
0
votes
0answers
5 views

Double DQN Agent can't complete environment outside of training

I have made my first custom OpenAi environment and have to build a double DQN agent to learn about it. Currently, this agent can get through training and finish episodes without any problems, please ...
1
vote
0answers
24 views

Whether the correlation structure of random field $Z(u) + Y(u)$ is equal to the correlation structure of $Z(u)$ plus that of $Y(u)$?

I want to simulate a Gaussian random field (RF) with correlation structure (represented by the geostatistic tool 'semivariogram' $\gamma (h) \: +\: pure \: nugget \: effect$). I want to know whether ...
1
vote
0answers
36 views

Categorical Cross Entropy Loss Derivation

I understand the categorical cross entropy based loss function to be the following. $$J(w) = \sum_{i=1}^ny_i\ln[P(y_i|x_i,w)]$$ where $$\ln\left[P(y_i|x_i,w)\right] = \sum_{i=k}^Kr_{ik}P(y_i=k|x_i,w)$$...
0
votes
0answers
16 views

LSTM forecasting and multicollinearity [closed]

I'm currently developing an LSTM RNN model to do multivariate time series forecasting. I have the time series with target feature $y_t = (y_1,y_2, \dotsc , y_T)$ with input features $\bf{X}_t = \bf{X}...
1
vote
0answers
7 views

How to tune LightGBM parameters to overcome underfitting? [closed]

I'm using LightGBM for a regression task. My training data's shape is (2000000, 1600), which means the number of training data is 2 million +, and each sample has 1600 features. The figure below is ...
0
votes
0answers
21 views

Do we really need to drop first in one-hot encoding? [duplicate]

Is there any consensus over whether one needs to drop first when performing one-hot encoding. With reference to here and here, I am thinking that only when you are using OLS then you need to leave one ...
1
vote
1answer
19 views

Are there any statistical properties from training data that imply whether feature selection will help improve an ML model's performance?

I was just wondering if there are any statistical tests/values we can look over the training data to know if applying feature selection can improve a model's performance when training on the data (...
0
votes
0answers
9 views

Is it possible to know whether a linear SVM is overfitting from the features' weight and value distribution in training?

I have a text sentiment classification model trained using linear SVM on 2500 training instances with around 14000 features(word), every sample is represented as binary vector with 1 indicate presence ...
0
votes
0answers
11 views

Assigning higher weights to more recent observations in a Neural Network

I have time-series data and am using a Neural Network for the purposes of forecasting forward. I have 20 years of monthly data but would like to assign a higher weight to those observations that have ...
4
votes
2answers
231 views

Neural network to read short strings - translational invariance in CNNs

I have a series of short strings that each describe some item (one item per string). The people who write these strings can get pretty creative when it comes to spelling. For each string, I also have ...
7
votes
1answer
81 views
+50

Difference between distribution shift, data shift, concept drift and model drift

Lately, I am seeing both terms used interchangeably in several scenarios. Joaquin Quiñonero in MIT press (NIPS), Dataset Shift in ML NIPS 2021 workshop in DistShift Model drift: Towards Data Science ...
1
vote
0answers
13 views

Rademacher Complexity of the family of constant-valued functions

Let $\mathcal{H}$ be a family of constant-valued functions with values in the closed interval $[a, b]$,how to calculate the rademacher complexity of $\mathcal{H}$? We know that the definition of ...
0
votes
0answers
15 views

How to generate adversarial examples by using random Gaussian noise as input belonging to certain class ( Targeted adversarial learning)

I have a image dataset with two classes: [0,1] and a trained model able to classify these two classes. Now, I want to generate an adversarial example belonging to a certain class, (say 0) by using ...
0
votes
0answers
11 views

Feature selection in multivariate time series forecasting

I'm currently developing a model (LSTM) to do forecasting, which has a large number of possible predictors. I have briefly searched for dimensionality reduction and feature engineering techniques for ...
0
votes
0answers
5 views

Correct Loss for bounded Multiple Regression

Suppose I have target vector y = (y_1, y_2, ..., y_n) where y_i in [0, inf) for all ...
0
votes
1answer
21 views

Do I refit a new model if I drop insignificant variables from the prior model?

I'm working on building a prediction model. I used group LASSO to perform some variable selection and ended up with a model that performs quite well. However, there are about 100 inputs right now and ...
-1
votes
1answer
67 views

Proper approach for image recognition of ~1000 symbols

We have a dataset of black symbols in grey squares (like attached below). The symbols are various letters (arabic, greek) as well as numbers in many distinct fonts; altogether ~1000 different images. ...
1
vote
0answers
24 views

Why to calculate $\mathbf{weighted}$ average of the leaf node impurities in decision trees? Why not to just add entropies up without weights?

In decision trees why do we calculate weighted average of entropies of each leaf when we calculate the entropy of target variable given some feature? The question is "Why is it weighted average? ...
2
votes
1answer
48 views

Formal treatment of overfitting on the test set

Assume that I split randomly the data into training and test sets. Suppose that I build a machine learning model using the training set. And suppose that I evaluate the accuracy of the model on the ...
0
votes
0answers
23 views

Can we accumulate deterministic and probabilistic approaches for regression and classification problems?

I am trying to accumulate various methods of regression and classification problems into two major approaches (specially in parametric form): Probabilistic: Here, we estimate the hypothesis function ...
0
votes
1answer
29 views

connection between multi-label classification and multi-class classification

For a dataset with multi-label judgement, e.g., coco dataset but where we only want to predict the most-possible label. There're multiple ways, for example : 1) train as a multi-label learning(each ...
0
votes
0answers
13 views

Information Value is Over 1

I'm using weight of evidence and calculated the information value using a library. For some of the variables, the IV is over 1 or even infinite. I'm not too sure what that's supposed to mean, since ...

1
2 3 4 5
352