2

How to develop a program that can sort images by similarity is similar to the problem of how to develop a program that can sort words by how similar they look. For example: "theory" is more similar to "teoryyy" than to "abc". What determines similarity of two words or images are these factors: how many parts are common to both images how many parts are ...


2

The Hough Transform extended to orthogonal ellipses uses this model, accumulating on $\theta$ for all $\{x, y\}$ with parameter matrix \begin{Bmatrix} c_x & c_y \\ r_x & r_y \end{Bmatrix} where $$1 = \dfrac {(x - c_x) \, \cos \theta} {r_x} + \dfrac {(y - c_y) \, \sin \theta} {r_y}$$ The question is looking to detect the normal lines, so any of ...


2

$l_{2,1}$ is a matrix norm, as stated in this paper. For a certain matrix $A \in \mathbb{R}^{r\times c}$, we have $$||A||_{2,1} = \sum_{i=1}^r \sqrt{\sum_{j=1}^c A_{ij}^2}$$ You first apply $l_2$ norm along the columns to obtain a vector with r dimensions. Then, you apply $l_1$ norm to that vector to obtain a real number. You can generalize this notation ...


2

Consider a dataset $S \in \mathbb{R}^{N \times (M + 1)}$ with $N$ observations (or examples), where each observation $S_i \in \mathbb{R}^{M + 1}$ is composed of $M$ elements, one value for each of the $M$ features (or independent variables), $f_1, \dots f_M$, and the corresponding target value $t_i$. A decision tree algorithm (DTA), such as the ID3 ...


2

For extra input that does not matter, you should not input it to the network. Feature selection, the process of finding and selecting the most useful features in a dataset, is a crucial step of the machine learning pipeline. Unnecessary features decrease training speed, decrease model interpretability, and, most importantly, decrease ...


2

This should be possible, considering universal approximation theorem you should be able to build a ann that approximates features that gives the most likely best feature set for a different net to train on. I would us a rnn for with a softmax output layer that ranks features by performance. You can find a good explanation of softmax here: https://developers....


2

Though there is no universal method which can be blindly used for all datasets, but here is what i usually do; Fill missing values using interpolation or mean, if missing values are less than 10-15 percent of number of rows else drop the column. Encode categorical data using some kind of encoding, e.g. one hot, etc. Then normalize/rescale columns. Now look ...


2

manual feature engineering started becoming obsolete That is wrong. Any suggestion on when to use manual feature engineering, feature learning or a combination of the two? Deep learning is awesome for natural signals like images, audio or large amounts of unstructured text (e.g. arbitrary crawled websites) There are some basic steps that make almost ...


2

Additional features can also cause overfitting if they have low or misleading information. Consider the following problem: $X = [1, 3, 3, 4, 5]$, $Y = [1, 3, 4, 4, 5]$. Suppose that the real dataset was generated from the relationship: $Y = X$, with a probability of 0.2 of adding or subtracting 1. A reasonable model estimate is $Y = X$. Note that no ...


2

Automated feature engineering, if it is part of any aproach towards general intelligence, cannot be the whole solution. The search for features that are meaningful, as opposed to those that simply exist with no utility, needs some guidance. In machine learning, feature engineering is typically a search for features that improve performance at a specific task,...


1

One way you can definitely approach the problem is by using (Deep) Reinforcement Learning (DRL). YouTube is actually using DRL as well to suggest videos to users in order to maximize users' engagement with their website. For more information (and further references to papers explaining how other major companies implement their recommendation systems), see ...


1

SVM complexity is $O(\max(n,d)\min(n,d)^2)$ according to Chapelle, Olivier. "Training a support vector machine in the primal." Neural Computation 19.5 (2007): 1155-1178. $n$ is the number of instances and $d$ is the number of dimensions. I'm assuming that you have more instances than dimensions giving a complexity of $O(nd^2)$. Hopefully this ...


1

Now I want to check if I can predict B directly from A, since, in my understanding, this would mean that info on B is already inside A. This will help inform you how much redundancy there is between A and B. However, even if you can predict B with 100% accuracy from A, you may still be better off using A+B (instead of A alone) to predict C. If I get good ...


1

Since you have all your data in a table, a relatively simple thing to do is to consider each column independently, and then seeing if the output variable (cost incurred) has a correlation to that. If the column has no (or very low correlation) with the output variable, then consider it to be not important. The ones that make the cut are then considered ...


1

It depends on the used network as well as the feeding mechanism but let's give an example; When working with LSTM, giving the time data (as an integer sequence) in addition to the time-series data(coming from features) dramatically increases the performance of the network. [$X_{0}$,$X_{1}$, ...] $\rightarrow$ [[$X_{0}$,$t_{0}$],$[X_{1}$,$t_{1}$], ...] ...


1

You don't need to re-train on the fly. What you're looking for is an embedded feature selection algorithm, and even more precisely, one that minimizes the number of responses required. I think this might be one of the rare cases where genetic and evolutionary approaches are the obviously correct choice. Genetic Programming is a technique for finding models ...


1

You may be mixing the concept of features with data granularity. Scan data after post processing is usually a cube of discrete density averages indexed by the three dimensions $x$, $y$, and $z$ corresponding to the standard directional references used in medicine. Features refer to what is extracted from those density readings, indicating concentrated ...


1

Have you tried any possibilities? You have many possible approaches. FeatureMiner or SkLearn are popular options for your purposes. The documentation from these two possibilities is also interesting.


Only top voted, non community-wiki answers of a minimum length are eligible