Wiki

An activation function is a function that runs on a neuron in a neural network and is responsible for mapping the neuron's input to its output.

Global optimization is a branch of applied mathematics and numerical analysis that attempts to find the minimum or maximum of a function on a given set. It is often described as a minimization problem, because the maximization of real-valued functions can be derived by analogy with minimization. The difference between global optimization and local optimization is that the former focuses on finding the extreme value on a given set, […]

Feedforward neural network is a relatively simple artificial neural network. Its internal parameters are propagated unidirectionally from the input layer to the output layer. Unlike recursive neural network, it does not form a directed loop internally. Feedforward is also called forward. From the perspective of signal flow, after the input signal enters the network, the signal flow is unidirectional […]

Sampling is a commonly used inferential statistical method. It refers to extracting a part of individuals from the target population (Population, or parent population) as a sample. By observing one or some attributes of the sample, an estimate with a certain reliability of the quantitative characteristics of the population is obtained based on the data obtained, thereby achieving an understanding of the population.

Sentiment analysis is a common method in natural language processing. It is based on the vocabulary analysis of text and determines the specific emotions contained in it. Sentiment analysis is similar to sentiment analysis, but sentiment analysis contains more types of emotional information. The sentiment dictionary officially released by the National Research Council of Canada contains the following 8 emotions: Angry […]

The frequency school believes that the world is deterministic and there is an ontology with unchanging truth value. The goal of the frequency school is to find the truth value or its range. The view of the frequency school is that there must be some deep generation mechanism behind random events. Although the event itself is random, this mechanism is deterministic. Different from the frequency school is the Bayesian school. The former […]

The training error is the error that occurs during data training. It is the average loss of the model on the training data.

Maximum expectation is an algorithm for finding the maximum likelihood estimate and maximum a posteriori estimate of parameters in a probabilistic model, where the probabilistic model is based on unobservable dependent variables. The maximum expectation algorithm is often used in the field of data clustering in machine learning and computer vision. It is calculated in two steps: Calculate the expectation E: Use the existing estimates of the hidden variables […]

Overfitting is a phenomenon in machine learning. It refers to the situation where some attributes in the sample that are not needed for classification are learned. In this case, the learned decision tree model is not the optimal model and will lead to a decrease in generalization performance.

Expected loss is the ability to predict all samples, which is a global concept; empirical risk is a local concept, which only represents the ability of the decision function to predict samples in the training data set. Empirical risk and expected risk Empirical risk is local, based on the minimization of the loss function of all sample points in the training set, and the empirical risk is locally optimal and can be realistically obtained; […]

Naive Bayes Classifier (NBC) is a conditional probability classifier based on Naive Bayes.

Naive Bayes is a classification algorithm based on probability theory that predicts and classifies only based on the probability of each category. This algorithm is based on the Bayes formula.

Paired t-test is a commonly used t-test. It refers to analyzing two groups of samples from the same population under different conditions to evaluate whether the different conditions have a significant impact. Different conditions can refer to different storage environments, different measurement systems, etc.

Underfitting refers to the situation where the model does not fit the training data well. It is usually used to evaluate the learning and generalization ability of the model.

The definition of a classifier is to construct a classification model based on existing data. The model can map the data records in the database to one of the given categories, so as to be applied to data prediction. The construction and implementation of the classifier generally go through the following steps: Select samples (including positive samples and negative samples […]

Weight is a relative concept, referring to a certain indicator. The weight of an indicator refers to the relative importance of the indicator in the overall evaluation.

Undersampling is a method to alleviate class imbalance by discarding some samples, that is, appropriately undersampling the categories with a large number of samples (majority classes) in the training set.

Soft margin is a method used to deal with linear inseparable problems and reduce the impact of noise. Soft margin is a method that allows some errors to exist during classification.

The radial basis function (RBF) is a scalar function that is symmetric along the radial direction. It is usually defined as the distance from any point X in space to a certain center X._c It can be written as K ( || X – X _c || ), its effect is often local, that is, when X is far away from X_c The function value is very small.

Quantum computing is a new type of computing based on quantum effects. The basic principle is to use quantum bits as information encoding and storage units, and to complete computing tasks through the controlled evolution of a large number of quantum bits. Comparison between quantum computing and traditional computing (1) Information expression In traditional computing, the unit of computer operation is a ratio of 0 or 1 […]

A quantum computer is a device that uses quantum logic to perform general computations. It is a specific implementation of quantum computing.

Quantum neural network (QNN) is a network composed of several quantum neurons according to a certain topological structure.

Robustness refers to the ability of a computer system to handle errors during execution and the ability of an algorithm to continue to operate normally when encountering anomalies such as input and calculation.

Supervised learning is a machine learning method in which the output is related to the input. A pattern can be learned or established from the training data, and new instances can be inferred based on this pattern.