5

This is conditioning in the sense of conditional probability. The idea is that the authors have some "standard physically-inspired features". They are splitting the data up into bins based on the values of these features, and then training a model for each bin. They are then examining the differences between the models. Usually this is done to learn ...


5

Non-correlation does not imply independence, that is, if two features are not correlated (i.e. zero correlation), it does not mean that they are independent. But (non-zero) correlation implies dependence (see https://stats.stackexchange.com/q/113417/82135 for more details). So, if you have non-zero correlation between two features, it means they are ...


3

Introduction Bag-of-features (BoF) (also known as bag-of-visual-words) is a method to represent the features of images (i.e. a feature extraction/generation/representation algorithm). BoF is inspired by the bag-of-words model often used in the context of NLP, hence the name. In the context of computer vision, BoF can be used for different purposes, such as ...


3

Data pre-processing and feature extraction are by far the most important part of any machine learning algorithm. It's even more important that the model you choose to do the classification. Unfortunately, pre-processing and feature extraction are completely different for each type of data. You need to play around with the data yourself to find out what ...


2

In general, the expression "temporal feature" might refer to any feature that is associated with or changes over time. However, in the context of signal processing, a temporal feature might refer to any feature of the data before being transformed to the Fourier, frequency or spectral domain, using the Fourier transform. In this context, the domain ...


2

As stated in your example, the three features are: an image, a price, a rating. Now, you want to build a model that uses all of these features and the simplest way to do is to feed them directly into the neural network, but it's inefficient and fundamentally flawed, due to the following reasons: In the first dense layer, the neural network will try to ...


2

The principal components (eigenvectors) correspond to the direction (in the original n-dimensional space) with the greatest variance in the data. The corresponding eigenvalue is a number that indicates how much variance there is in the data along that eigenvector (or principal component). Thus, feature 2 is the most important (based on ...


2

Simply said, there is no specific "meaning" to the features generated. They are simply features that are fitted through math and calculus, and nobody knows what they represent exactly, and will never knows. However we can run PCA (Principal Component Analysis) to see which feature is the most "important" of all, aka which feature affects the most in the ...


2

We do it experimentally; you're able to look at what each layer is learning by tweaking various values throughout the network and doing gradient ascent. For more detail, watch this lecture: https://www.youtube.com/watch?v=6wcs6szJWMY&list=PL3FW7Lu3i5JvHM8ljYj-zLfQRF3EO8sYv&index=12 it provides many methods used for understanding exactly what your ...


2

If all you features are binary, then, you don't need to apply normalization on them. Since their values are on the same scale already.


2

In general terms yes. Because what the ML algorithms do in general is to learn the hidden probability density function of the target examples (cats, dogs..). And that is done by learning the conditional probability function between inputs, $X$, and target outputs, $y$, for discriminative models or by learning the joint probability function for generative ...


2

Yes, neural networks learn features themselves freeing you from the need to manually engineer them. I will illustrate it here with a toy problem. Let's assume that we want to learn the areas of parallelograms built on pairs of vectors: The input data are six coordinates: $(x_1, y_1, x_2, y_2, x_3, y_3)$. import numpy as np n_tr = 1000 # training data x_tr = ...


2

Feature engineering may be necessary when one cannot achieve acceptable error rate — within a budget or in principle. NN may be stalling due to information bottleneck: too many pigeons, not enough holes. In that case, custom features may provide slightly better information compression. (Alas, this is not a panacea: some layer(s) may still be too narrow. That'...


2

No, channels do not have to only represent colours. It is common for them to represent other things, even without considering feature maps. For instance RGBD images, where D is a depth measurement or distance from a sensor. Or when CNNs are applied to grid-based games, such as chess or go with AlphaZero, where the input channels are information about game ...


1

You can find a brief explanation of hierarchical feature selection in the following from "An Empirical Evaluation of Hierarchical Feature Selection Methods for Classification in Bioinformatics Datasets with Gene Ontology-based Features" paper: Hierarchical feature selection is a new research area in machine learning/data mining, which consists of ...


1

Well, I suppose one can use some kind of contrastive learning in this case. A famous example of the establishment of relation between two different representations is the CLIP - Contrastive Language–Image Pre-training, where model gets a huge corpus of image captions and images and the image caption is passed through the language model, and the image itself ...


1

The network architecture is relevant to this question. Convolutional neural network architectures enforce the building up of features because the neurons in earlier layers have access to a small number of input pixels. Neurons in deeper layers are connected (indirectly) to more and more pixels, so it makes sense that they identify larger and larger features. ...


1

No, it does not take into account the curvature. But, if curvature is important for you, then, it would be a good idea to look at Ricci flow and its applications in neural networks.


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

The body of your post seems to be asking a completely separate question than the title of your post, so I will answer both: "Body: How do I complete the goal of this program?" Your dataset does not have the dependent variable, which is the outcome of the game (win/loss/draw). What I am assuming is that you have a way of looking up the outcome of ...


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

Pragmatically you could use the discriminatory from a GAN for outlier detection. Ideally you'd start collecting fakes now and do a normal model on both good and bad cases. In the absence of that you can create a GAN to create realistic looking fakes on only real cases and then take the discriminator from that GAN to flag real-life cases for manual checks. ...


1

The problem which you have is a classification problem. You assume a class "good users" and a distinct class "bad users". You want to train an AI to tell the two apart, but all your examples are "good users". Any reasonable AI will draw the logical conclusion from those examples: all users are good users. That's a 100% match for the training data.


1

I would refer to your problem as having a continuous state space. By using a 32-bit float variable you discritize it. However, creating states for every possible value of a 32-bit float variable is probably too much. You should decide on: the variable range: what is the real range of the position variables (e.g. from 0 m to 10 m), and what is the ...


1

Two popular methods I’ve seen done: 1) For each feature, remove it and run the model and see the impact it has on the result. The idea is that the larger the impact, the more pertinent it was to the result. 2) Look at the gradients magnitude $|\nabla_f {y} |$. You can either look at the raw gradient or look at the guided back-propagation which is just ...


1

Each feature map (or kernel) is independent of each other. If you had $3$ of these filters, your output shape would be $(28, 28, 3)$ (given the appropriate amount of padding and stride) with a total of $75*3=225$ trainable weights.


1

I don't know what kind of price data you're dealing with. I suppose the order of the data matters a lot, so my suggestion would be: Use LSTM as it handles time series better You can predict 3 consecutive numbers from an RNN as the next three days' predictions Try regression first, it is likely it will not work (or just flatten the curves, depend on your ...


1

In short ANNs don't have problems with "different types" of data as long as they are represented using real numbers: the inputs for your ANN represent lengths and are easy to understand and process. The variable number of inputs is a little bit more tricky. In general, it is not a problem either. The net will compensate for the absence of some ...


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