Wiki

ODE is the most commonly used strategy for semi-naive Bayes classifiers. ODE assumes that each attribute depends on at most one other attribute outside the category.

The polynomial kernel function refers to a kernel function expressed in polynomial form. It is a non-standard kernel function suitable for orthogonal normalized data. Its specific form is shown in the figure.

The principle of multiple interpretations is the idea that all hypotheses that are consistent with empirical observations should be retained.

Hyperplane partitioning means that if two disjoint convex sets are both open, then there exists a hyperplane that can separate them.

Stratified sampling is a sampling method that involves stratification before sampling. It is a commonly used sampling method in statistics.

Symbolic learning refers to machine learning methods that functionally simulate human learning abilities.

Symbolism is a school of thought in the field of artificial intelligence that believes in mathematical logic.

The unit step function is also called the Heaviside step function and is defined as follows: Its graph is as follows: Related terms: impulse function,

Von Neumann architecture is a computer design concept that combines program instruction memory and data memory.

Secondary learning refers to repeated learning when the first learning result is not ideal.

Unequal costs refer to situations where different costs are assigned to the losses incurred by each category.

Unsaturated games are inspired by heuristic methods rather than theoretical analysis.

Adversarial network is an implementation of generative adversarial network, which is used to generate adversarial samples in batches for a specified neural network model.

Adversarial samples refer to inputs that cause the network to output incorrectly in a neural network. The input samples are formed by deliberately adding slight interference to the data set. The input after interference causes the model to give incorrect outputs with high confidence. The input samples are adversarial samples. This behavior is usually regarded as an adversarial attack on the neural network model. […]

Affine layers are fully connected layers in neural networks, where affine can be seen as interconnections between neurons in different layers, and in many ways can be seen as a "standard" layer of a neural network. The general form of an affine layer is as follows y = f(wx + b) Note: x is the layer input, w is a parameter, and b is a bias […]

Metric learning can also be considered as similarity. Metric learning is to measure the similarity between samples, which is one of the core issues of pattern recognition. The goal of metric learning is to minimize the distance between samples of the same type and maximize the distance between samples of different types.

Multi-classification, sometimes called multi-class classification, refers to the classification of more than two categories in a classification task. Existing multi-class classification techniques can be divided into (i) conversion to binary (ii) expansion from binary and (iii) hierarchical classification. Common strategies 1) One-vs.-all strategy requires establishing a unique […]

Multilayer Perceptron (MLP) is a forward-structured artificial neural network that maps a set of input vectors to a set of output vectors. It can be viewed as a directed graph consisting of multiple layers of nodes, each of which is fully connected to the next layer. In addition to the input node, each […]

Modality refers to the specific way people receive information. Since multimedia data is often a medium for transmitting multiple types of information (for example, a video often transmits text information, visual information, and auditory information at the same time), multimodal learning has gradually developed into the main means of analyzing and understanding multimedia content. Multimodal learning mainly includes the following […]

The generalization error upper bound refers to the maximum value allowed for the generalization error. Exceeding this upper bound will affect the feasibility of machine learning. Generalization error refers to the error generated in the process of generalizing from the training set to outside the training set. It is usually obtained by subtracting the training error from the error outside the training set, that is, the error expectation on the entire input space. Because the error upper bound has a wide range of […]

Multidimensional scaling (MDS) is a visualization of the distances between a set of objects, and can also be used as an unsupervised dimensionality reduction algorithm. It is a dimensionality reduction method that can alleviate the sparse sample data and difficulty in distance calculation that occur in high-dimensional situations. It is a linear dimensionality reduction method that is different from both principal component analysis and linear dimensionality reduction analysis […]

Multiple linear regression is a linear regression for multiple variables. The multiple linear regression method is similar to univariate regression, except that there are more independent variables and parameters. Common functions of multiple regression Linear correlation coefficient between variables cor(dataframe) Scatter plot matrix scatterplotMatrix […]

Occam's razor means that if there are multiple hypotheses that are consistent with observations, the simplest one should be chosen. Occam's razor is often used as a heuristic technique, a tool to help people develop theoretical models, and cannot be used as a basis for judging theories.

Out-of-bag estimates refer to test results where the samples used for testing do not appear in the training set.