Questions tagged [machine-learning]
How can we build computer systems that automatically improve with experience, and what are the fundamental laws that govern all learning processes?
3,386 questions
0 votes
0 answers
26 views
When should ensemble learning be used and when is better performance than a single model guaranteed? [closed]
I have a question about ensemble learning methods. When should ensemble learning be used and when is better performance than a single model guaranteed? More specifically: Are there theoretical ...
- 1
0 votes
0 answers
21 views
Optimal lags in neuralprophet model [closed]
I am using neuralprophet model in time series data having 2 columns( ds and y). ds is timestamp and y is numerical column. As I am using hyperparameter tuning , so how to select optimal n_lags value ...
0 votes
0 answers
22 views
Verifying One-vs-All Precision and Recall calculations from a multi-class confusion matrix
I am studying multi-class classification metrics and want to confirm the correct way to compute them from a confusion matrix. A weather classifier labels days as Sunny, Rainy, Cloudy. The test results ...
- 1
0 votes
0 answers
29 views
Notation Convention in Linear Models
Notation Convention in Linear Models: Why $\theta^\top x$ instead of $\theta x$? Question: I'm working with CMU 10-414 Lecture 2 and I'm curious about the notation convention used to represent the ...
- 107
3 votes
0 answers
138 views
How to quickly come back up to speed on math I used 50+ years ago? [closed]
I'm 72, just got laid off, and am going back to school to learn machine learning. This is going to require higher maths than I took in 1971, which was Calculus I. I'm finding that mental reflexes ...
- 39
2 votes
1 answer
106 views
Integration by parts from perceptron capacity calculation
I am working through Chapter 6 of the book Statistical Mechanics of Machine Learning by Engel and Van den Broeck. I am stuck on the following integral, going from line 6.13 to line 6.14 of the book. I ...
- 100
1 vote
0 answers
35 views
4th Order Tensor multiplication Rules for Sparse Regression analysis
I am working on a problem which involves working with stress and deformation tensors of the order 4. I have a set of data at different time steps for 20 cases and each element stress is 3x3 matrix, so ...
1 vote
1 answer
43 views
Does the union of two datasets form a mixture distribution? [closed]
I have two datasets: $A := \{X_i\}_{i=1}^{n_a}$ sampled from distribution $P_A$, and $B := \{X_j\}_{j=1}^{n_b}$ sampled from distribution $P_B$. Let $n = n_a + n_b$ be the total sample size, and ...
- 11
1 vote
0 answers
25 views
Understanding BottleNeck Block in ResNet [closed]
I’m new to machine learning and trying to strengthen my understanding and coding skills for neural networks. Recently, I was exploring the ResNet architecture and found this article really helpful: ...
- 11
1 vote
0 answers
32 views
Why is X following $\mathcal{N}(\mu + \Lambda z, \Phi)$ in the Factor Analysis model?
I’m working through some notes on Factor Analysis and I noticed something that confused me. We have $ X = \mu + \Lambda z + \epsilon $ with $z \sim \mathcal{N}(0,I_s)$, $\epsilon \sim \mathcal{N}(0,\...
0 votes
0 answers
15 views
Eigenvalues and eigenfunctions of periodic kernel
The periodic kernel which is sometimes used in Gaussian process regression models looks like $$ k(x,x') = \sigma^2 \exp \left(-\frac{2 \sin^2(\pi|x-x'|/p)}{\ell^2}\right) $$ for parameters $\sigma^2$, ...
- 71
2 votes
2 answers
222 views
Machine learning: what is the proper name for derivative of a function against a matrix?
In machine learning, it is typical to see a so-called weight matrix. As a low-dimensional example, let this matrix be defined as, $$W = \begin{bmatrix} w_{11} & w_{12} \\\ w_{21} & w_{22} \end{...
- 8,711
1 vote
0 answers
56 views
Jensen's inequality vs concavity of $\log(f(x))$: why EM lower bound works for GMM?
I have a doubt about the relationship between Jensen's inequality and the concavity of a composite function, specifically $\log(f(x))$. Let $f: \mathbb{R}^n \to \mathbb{R}_{>0}$ be a positive ...
- 163
0 votes
1 answer
45 views
Decision boundary in linear classification models
The decision boundary, $y = \mathbf{w}^T\mathbf{x} + b = 0$, is the decision boundary in linear classification models. When $\mathbf{x} \in \mathbb{R}^2$ and $\mathbf{w} \in \mathbb{R}^2$, then $y \in ...
- 129
3 votes
1 answer
100 views
Linear model classification
I am trying to understand the geometry for linear classification models. A linear model, according to Bishop's books, is defined as: $\mathbf{y} = \mathbf{w}^T \mathbf{x} + w_0$. For instace we have ...
- 129