Approximate Bayesian Computation (ABC) is a computational method based on Bayesian statistics that can be used to estimate the posterior distribution of model parameters.

Approximate inference methods refer to sampling and learning from a large amount of data and using hypothesis-verification logic to continuously approach the true model.

In mathematics, a distance matrix is a matrix (i.e. a two-dimensional array) containing the distances between pairs of points.

Plug and Play Generative Network (PPGN) is a model proposed by Nguyen et al. in 2016.

The column name attribute refers to the 'name-related' feature of the data, and the corresponding value is the name of some symbol or thing.

Cumulative error back propagation is a neural network algorithm that uses a gradient descent-based strategy to adjust parameters in the negative gradient direction of the target, with the goal of minimizing the training error. It is also called the "back propagation algorithm", or "BP algorithm" for short.

Grouping related samples together is generally used for unsupervised learning. Once all samples are grouped, researchers can optionally assign meaning to each cluster. There are many clustering algorithms, for example, the k-means algorithm clusters samples based on their proximity to the centroid, as shown below: After that, researchers can […]

The majority voting method is a voting method that requires more than half of the valid votes to be recognized. When multiple classifiers predict a certain category, only the part that is higher than half of the total results will be predicted. The following is the formula for representation: $latex {H{ \left( {x} \right) }\text{ […]

Manifold learning is a basic method in pattern recognition. It seeks the essence of things and the inherent laws of data generation based on observed phenomena. Manifold learning is divided into two types: linear manifold learning algorithm and nonlinear manifold learning algorithm. Nonlinear manifold learning algorithm includes isomap, Laplace eigenmap, and L[…]

The mean square error is the expected value that reflects the difference between the estimated value and the true value. It is often used to evaluate the degree of change in data and predict the accuracy of data. Assuming that there is a parameter , and its estimation function is , then $latex {MSE [...]

Machine translation is the use of computers to convert between different languages, usually translating from a source language into a target language. Translation Process From the perspective of human translation, the translation process can be broken down into the following: Decipher the meaning of the source text and recompile the meaning obtained after the analysis into the target language. Translation Methods General steps of machine translation […]

The manifold hypothesis is a common hypothesis in semi-supervised learning, and the other is the clustering hypothesis. The manifold hypothesis refers to examples with similar properties, which are usually in a small local area, so the labels are also very similar. This hypothesis reflects the local smoothness of the decision function. Unlike the clustering hypothesis that focuses on the overall characteristics, the manifold hypothesis focuses more on the model’s […]

Maximum likelihood estimation is an estimation method mainly used to determine model parameters by finding possible data that maximizes the model observation value and using it as the final choice. In maximum likelihood estimation, sampling satisfies the assumption of independent and identical distribution. Its purpose is to use known sample results to infer the parameter value that has the highest probability of leading to related results.

Lazy learning is a training set processing method that trains at the same time as receiving the test sample. In contrast, eager learning starts learning samples during the training phase. If the task data changes frequently, lazy learning can be used. No training is performed first, and the prediction request is received before the current data is estimated. […]

Analogy learning is a method of cognitive thinking and inference. It compares two types of things or situations to find their similarities at the object level, and then compares the relationship between things and situations based on this, and then obtains the corresponding solution by appropriately sorting/exchanging them to another thing. There are several classification methods for analogy learning […]

The holdout method is a model evaluation method that divides the dataset D into two mutually exclusive sets. Assuming that one set is the training set S and the other is the test set T, then: D = S ∪ T , S ∩ T = ∅ The division of the training/test set should keep the data distribution as consistent as possible. To avoid […]

Pruning is a method to stop a decision tree from branching. It is a method to solve the problem of overfitting in decision trees.

Hypothesis testing is a method of testing statistical assumptions, mainly used in inferential statistics. "Statistical assumptions" are a kind of scientific hypothesis testing, which is mainly carried out by observing the model of random variables. Under the premise that the unknown parameters can be estimated, appropriate inferences can be made about the unknown parameter values based on the results. The assumption of parameters in statistics is the assumption of one or more […]

Ensemble learning is the idea of combining multiple models into a high-precision model. It is mainly used in the field of machine learning. It is not a single machine learning algorithm, but a learning task completed by building and combining multiple learners. Ensemble learning can be used for classification problems, regression problems, feature selection, outlier detection, etc. It can be said that all machine learning […]

The error-correcting output coding method ECOC can convert multi-class problems into multiple two-class problems, and the error-correcting output code itself has error correction capabilities, which can improve the prediction accuracy of supervised learning algorithms. Output category encoding can convert multi-class problems into two categories, that is, each category corresponds to a binary bit string of length n, forming a total of m codewords, which […]

Empirical risk shows the model's ability to predict training samples. It is obtained by calculating the loss function once for all training samples and then accumulating and averaging. The loss function is the basis of expected risk, empirical risk, and structural risk. The loss function is for a single specific sample and represents the gap between the model's predicted value and the true value. […]

K-means clustering is a vector quantization method that was used in signal processing in the early days. It is currently mainly used as a clustering analysis method in the field of data mining. The purpose of k-means clustering is to divide n points into k clusters so that each point belongs to the cluster corresponding to the nearest mean, and use this as the clustering criterion. This type of problem […]

Margin theory is a concept in support vector machines, where the margin refers to the minimum distance between two types of samples divided by a hyperplane. Margin theory can be used to explain that when the training error of the AdaBoost algorithm is 0, continuing training can further improve the generalization performance of the model. Let x and y represent the input and […]

The perceptron is a binary linear classification model that can be regarded as the simplest form of a feedforward neural network, a model invented by Frank Rosenblatt in 1957. Its input is the feature vector of an instance, and its output is the category of the instance.

Approximate Bayesian Computation (ABC) is a computational method based on Bayesian statistics that can be used to estimate the posterior distribution of model parameters.

Approximate inference methods refer to sampling and learning from a large amount of data and using hypothesis-verification logic to continuously approach the true model.

In mathematics, a distance matrix is a matrix (i.e. a two-dimensional array) containing the distances between pairs of points.

Plug and Play Generative Network (PPGN) is a model proposed by Nguyen et al. in 2016.

The column name attribute refers to the 'name-related' feature of the data, and the corresponding value is the name of some symbol or thing.

Cumulative error back propagation is a neural network algorithm that uses a gradient descent-based strategy to adjust parameters in the negative gradient direction of the target, with the goal of minimizing the training error. It is also called the "back propagation algorithm", or "BP algorithm" for short.

Grouping related samples together is generally used for unsupervised learning. Once all samples are grouped, researchers can optionally assign meaning to each cluster. There are many clustering algorithms, for example, the k-means algorithm clusters samples based on their proximity to the centroid, as shown below: After that, researchers can […]

The majority voting method is a voting method that requires more than half of the valid votes to be recognized. When multiple classifiers predict a certain category, only the part that is higher than half of the total results will be predicted. The following is the formula for representation: $latex {H{ \left( {x} \right) }\text{ […]

Manifold learning is a basic method in pattern recognition. It seeks the essence of things and the inherent laws of data generation based on observed phenomena. Manifold learning is divided into two types: linear manifold learning algorithm and nonlinear manifold learning algorithm. Nonlinear manifold learning algorithm includes isomap, Laplace eigenmap, and L[…]

The mean square error is the expected value that reflects the difference between the estimated value and the true value. It is often used to evaluate the degree of change in data and predict the accuracy of data. Assuming that there is a parameter , and its estimation function is , then $latex {MSE [...]

Machine translation is the use of computers to convert between different languages, usually translating from a source language into a target language. Translation Process From the perspective of human translation, the translation process can be broken down into the following: Decipher the meaning of the source text and recompile the meaning obtained after the analysis into the target language. Translation Methods General steps of machine translation […]

The manifold hypothesis is a common hypothesis in semi-supervised learning, and the other is the clustering hypothesis. The manifold hypothesis refers to examples with similar properties, which are usually in a small local area, so the labels are also very similar. This hypothesis reflects the local smoothness of the decision function. Unlike the clustering hypothesis that focuses on the overall characteristics, the manifold hypothesis focuses more on the model’s […]

Maximum likelihood estimation is an estimation method mainly used to determine model parameters by finding possible data that maximizes the model observation value and using it as the final choice. In maximum likelihood estimation, sampling satisfies the assumption of independent and identical distribution. Its purpose is to use known sample results to infer the parameter value that has the highest probability of leading to related results.

Lazy learning is a training set processing method that trains at the same time as receiving the test sample. In contrast, eager learning starts learning samples during the training phase. If the task data changes frequently, lazy learning can be used. No training is performed first, and the prediction request is received before the current data is estimated. […]

Analogy learning is a method of cognitive thinking and inference. It compares two types of things or situations to find their similarities at the object level, and then compares the relationship between things and situations based on this, and then obtains the corresponding solution by appropriately sorting/exchanging them to another thing. There are several classification methods for analogy learning […]

The holdout method is a model evaluation method that divides the dataset D into two mutually exclusive sets. Assuming that one set is the training set S and the other is the test set T, then: D = S ∪ T , S ∩ T = ∅ The division of the training/test set should keep the data distribution as consistent as possible. To avoid […]

Pruning is a method to stop a decision tree from branching. It is a method to solve the problem of overfitting in decision trees.

Hypothesis testing is a method of testing statistical assumptions, mainly used in inferential statistics. "Statistical assumptions" are a kind of scientific hypothesis testing, which is mainly carried out by observing the model of random variables. Under the premise that the unknown parameters can be estimated, appropriate inferences can be made about the unknown parameter values based on the results. The assumption of parameters in statistics is the assumption of one or more […]

Ensemble learning is the idea of combining multiple models into a high-precision model. It is mainly used in the field of machine learning. It is not a single machine learning algorithm, but a learning task completed by building and combining multiple learners. Ensemble learning can be used for classification problems, regression problems, feature selection, outlier detection, etc. It can be said that all machine learning […]

The error-correcting output coding method ECOC can convert multi-class problems into multiple two-class problems, and the error-correcting output code itself has error correction capabilities, which can improve the prediction accuracy of supervised learning algorithms. Output category encoding can convert multi-class problems into two categories, that is, each category corresponds to a binary bit string of length n, forming a total of m codewords, which […]

Empirical risk shows the model's ability to predict training samples. It is obtained by calculating the loss function once for all training samples and then accumulating and averaging. The loss function is the basis of expected risk, empirical risk, and structural risk. The loss function is for a single specific sample and represents the gap between the model's predicted value and the true value. […]

K-means clustering is a vector quantization method that was used in signal processing in the early days. It is currently mainly used as a clustering analysis method in the field of data mining. The purpose of k-means clustering is to divide n points into k clusters so that each point belongs to the cluster corresponding to the nearest mean, and use this as the clustering criterion. This type of problem […]

Margin theory is a concept in support vector machines, where the margin refers to the minimum distance between two types of samples divided by a hyperplane. Margin theory can be used to explain that when the training error of the AdaBoost algorithm is 0, continuing training can further improve the generalization performance of the model. Let x and y represent the input and […]

The perceptron is a binary linear classification model that can be regarded as the simplest form of a feedforward neural network, a model invented by Frank Rosenblatt in 1957. Its input is the feature vector of an instance, and its output is the category of the instance.