. NET Platform Open Source project Quick glance (13) or piece accord.net frame function introduction

The accord.net framework is encapsulated and further developed on the basis of the Aforge.net project. Because Aforge.net focuses more on the underlying and breadth, the accord.net framework focuses more on machine learning algorithms and solutions that provide computer video, audio, signal processing, and statistical applications. The project is written in the C # language, Project home:http://accord-framework.net/

Description: The article is just a basic introduction, the main content is the official translation of documents and presentations, some of the English expression of personal ability is limited, not too familiar, so direct copy of the original text, there is more accurate to know the name of the Chinese can remind some of me, very feeling. I will use this component to do some simple data mining and machine learning tasks, the process and code will be published in this blog, interested can be concerned.

NET Open source directory:"directory" of this blog other. NET open source project articles directory

The original address of this article:. NET Platform Open Source project Quick glance (12) or piece accord.net frame function introduction

1.basic functions and introduction

accord.net for . NET Applications provide statistical analysis, machine learning, image processing, computer vision-related algorithms. The Accord.net framework extends the aforge.net framework and provides some new features. At the same time . NET environment for scientific Computing provides a complete development environment. The framework is split into multiple assemblies that can be downloaded directly from the official website or used by NuGet. You can refer to the following links:Https://github.com/accord-net/framework/wiki

three functional modules of the 1.1 framework

accord.net The framework consists of three large functional modules. For science and technology, signal and image processing, support components. The following is a brief description of the namespaces and features of the 3 models. This is the main namespace introduction that allows you to get in touch and understand whether their functionality is what you want.

1.1.1 Scientific Calculation

Accord.math : Includes matrix extenders and a set of methods for numerical computation and decomposition of matrices, as well as some numerical optimization algorithms for constrained and non-constrained problems, as well as some special functions and other ancillary tools.

Accord.statistics : Includes statistical models and methods such as probability distribution, hypothesis testing, linear and logistic regression, Hidden Markov model, (hidden) conditional random domain, principal component analysis, partial least squares discriminant analysis, kernel method, and many other related techniques.

accord.machinelearning : Provides algorithms for machine learning applications including support vector machines, decision trees, naive Bayesian models, K-means clustering algorithms, Gaussian mixed models and general-purpose algorithms such as RANSAC, cross-validation, and grid search.

Accord.neuro : Includes a large number of neural network learning algorithms, such as the Levenberg-marquardt,parallel resilient Backpropagation,nguyen-widrow initialization algorithm, Deep belief networks and many other neural network-related algorithms. See the reference Help documentation for details.

1.1.2 Signal and image processing

accord.imaging : Includes feature point detectors (such as Harris, SURF, FAST and FREAK), image filters, image matching and image stitching methods, as well as some feature extractors.

Accord.audio: contains a number of machine learning and statistics applications that require processing, conversion filters, and methods for handling audio signals.

accord.vision: real-time face detection and tracking, as well as the general detection, tracking and conversion methods in the flow of people images, there are dynamic template matching tracker.

1.1.3 Support Components

Mainly for some of the above components to provide data display, drawing of the control, divided into the following namespaces:

Accord.controls: includes bar charts, scatter plots, and tabular data browsing common to scientific computing applications.

Accord.Controls.Imaging: The WinForm control, which includes an image for display and processing, contains a dialog box that allows you to quickly display an image.

Accord.Controls.Audio: a WinForm control that displays waveform and audio dependency information.

Accord.Controls.Vision: includes tracking of head, face and hand movements as well as other computer vision-related tasks WinForm controls.

1.2 Supported algorithms introduction

below will The main functional algorithms included in the Accord.net framework are described by category. The source is mainly the official website introduction, has carried on the simple translation and the collation.

1.2.1 categories (classification)

SVM (Support vector machine), logistic Regression (logistic regression), decision Trees (decision Tree), Neural Networks (neural network), deep learning Networks deep Neural network), Levenberg-marquardt with Bayesian regularization, Restricted Boltzmann machines (limited Boltzmann machine), Sequence Classification (sequence classification), Hidden Markov classifiers and Hidden Conditional random fields (hidden Markov classifier and stealth conditional random field).

1.2.2 Regression (Regression)

Multiple linear regression (multivariate linear regression-univariate variables), multivariate linear regression (multivariate linear regression-multiple independent variables), Polynomial regression ( Polynomial regression), logarithmic regression (logarithmic regression), logistic regression (logistic regression), multinomial logistic regression (polynomial logistic regression) (Softmax) and Generalized linear models (generalized linear model), l2-regularized l2-loss logistic regression, l2-regularized logistic regression, l1-re gularized Logistic regression, l2-regularized logistic regression in the dual form and regression support vector machines 。

1.2.3 Cluster (clustering)

K-means, K-modes, mean-shift (mean Drift), Gaussian Mixture Models (Gaussian mixture model), binary split (two-yuan split), deep belief Networks (deeper belief network), Restricted Boltzmann Machines (limited Boltzmann machine). Clustering algorithms can be applied to arbitrary data, including images, data tables, video, and audio.

1.2.4 Probability distribution (distributions)

Includes more than 40 distributions of parametric and nonparametric estimates. Some common distributions include normal distribution, Cauchy distribution, hypergeometric distribution, Poisson distribution, Bernoulli, and some special distributions such as Kolmogorov-smirnov, Nakagami, Weibull, and von-mises distributions. Also includes multivariate distributions such as multivariate normal distribution, multinomial, Independent, Joint and Mixture distributions.

1.2.5 hypothesis Test (hypothesis Tests)

Over 35 statistical hypothesis tests, including unidirectional and bidirectional ANOVA tests, nonparametric tests such as Kolmogorov-smirnov tests and signal testing in the media. Contingency table tests such as the Kappa Test,with variations for multiple tables, as well as the Bhapkar and Bowker tes Ts And the more traditional chi-square, Z, F, T and Wald tests.

1.2.6 Nuclear method (Kernel Methods)

Kernel support vector machines, multi-class and multi-label vector machines, sequence minimization, least squares learning, probabilistic learning. Including special methods for linear machines such as Liblinear ' s methods for linear coordinate descent, linear Newton Me Thod, probabilistic coordinate descent, probabilistic coordinate descent in the Dual, probabilistic Newton Method for L 1 and L2 machines in both the dual and primal formulations.

1.2.7 Image (Imaging)

Interest and feature point detectors such as Harris,freak,surf,fast. Grayscale symbiosis Matrix, Border following,bag-of-visual-words (BoW), ransac-based homography estimation, integral images, Haralick  Textural feature extraction, and dense descriptors such as histogram of oriented gradients (HOG) and Local Binary Pattern (LBP). Several image Filters for image processing applications such as difference of Gaussians, Gabor, Niblack and Sauvola thre Sholding. There are also several image filters that are often used in image processing.

1.2.8 audio Signal (voice and Signal)

Audio signals are loaded, parsed, saved, filtered, and converted, such as the application of audio filters in the spatial domain and frequency domain. WAV file, audio capture, time domain filter, high pass, low pass, wave rectifier filter. Frequency-domain operators such as differential rectification filter and comb filter with Dirac ' s delta functions. Signal generators for cosine, Impulse, Square signals.

1.2.9 Visual (Vision)

Real-time face detection and tracking, as well as image flow detection, tracking, conversion of the general detection method. Contains cascade Definitions, Camshift and Dynamic Template Matching trackers. Includes pre-created classifiers for human faces and some facial features such as noses.

1.3 Related Resources

from the project homepage: http://accord-framework.net/ Download the “ archive” compressed package, including almost all online resources. For example, introduce several key resources:

Debug is a set of assemblies for debugging, Docs is a help document, Externals is a number of auxiliary components, Release is a DLL assembly version for different. NET environments, samples is the case source code, Setup is the program installed, sources is the source code for the project, and unit tests is the units test code.

accord.net Framework source code hosted in GitHub:

https://github.com/accord-net/framework/

There are a number of introductory resources and tutorials, for example, to view the list bar toggle on the right side of the page:

Https://github.com/accord-net/framework/wiki/How-to-use

. NET Platform Open Source project Quick glance (13) or pieces accord.net Framework function Introduction