Debian Science Machine Learning packages

AutoClass résout le problème de la découverte automatique de classes dans des données (parfois appelé clustering ou unsupervised learning) à la différence de la génération de description de classes à partir d'exemples nommés (appelé supervised learning). Il essaie de découvrir des classes « naturelles » dans les données. AutoClass est applicable à des observations de choses qui peuvent être décrites par un ensemble d'attributs, sans se référer à d'autres éléments. Les valeurs de données correspondantes à chaque attribut sont limitées à être soit des nombres ou les éléments d'un ensemble fixé de symboles. Avec des données numériques, une erreur de mesure doit être fournie.

Caffe est un cadriciel d'apprentissage profond créé avec le souci de l'expression, la vitesse et la modularité. Il est développé par le BAIR («⋅Berkeley AI Research Lab⋅») et des contributeurs de la communauté.

Ce méta-paquet fournit la version CPU_ONLY de caffe⋅:

 –⋅caffe-tools-cpu ;
 –⋅libcaffe-cpu* ;
 –⋅python3-caffe-cpu.
et suggère les paquets suivants⋅:
 –⋅libcaffe-cpu-dev ;
 –⋅caffe-doc.

Attention, cette version CPU_ONLY ne peut pas coexister avec la version CUDA.

GNU Prolog is a free Prolog compiler with constraint solving over finite domains (FD). GNU Prolog is largely compliant with the ISO standard and is part of the Prolog Commons initiative.

This package contains the compiler and runtime system for the ISO standard version of GNU Prolog, including the prototype modules implementation.

Fast Artificial Neural Network Library is a free open source neural network library, which implements multilayer artificial neural networks in C with support for both fully connected and sparsely connected networks. Cross-platform execution in both fixed and floating point are supported. It includes a framework for easy handling of training data sets. It is easy to use, versatile, well documented, and fast.

This package contains the header files and static libraries which are needed for developing libfann applications.

GAlib contains a set of C++ genetic algorithm objects. The library includes tools for using genetic algorithms to do optimization in any C++ program using any representation and genetic operators. The documentation includes an extensive overview of how to implement a genetic algorithm as well as examples illustrating customizations to the GAlib classes.

This package contains the development files.

LIBLINEAR is a library for learning linear classifiers for large scale applications. It supports Support Vector Machines (SVM) with L2 and L1 loss, logistic regression, multi class classification and also Linear Programming Machines (L1-regularized SVMs). Its computational complexity scales linearly with the number of training examples making it one of the fastest SVM solvers around.

This package contains the header files and static libraries.

This package contains the mlpack Library development files.

Machine Learning Pack (mlpack) is an intuitive, fast, scalable C++ machine learning library, meant to be a machine learning analog to LAPACK. It aims to implement a wide array of machine learning methods and function as a "swiss army knife" for machine learning researchers.

This library implements Optimized Cutting Plane Algorithm (OCAS) for training linear Support Vector Machine (SVM) classifiers from large-scale data. The computational effort of OCAS scales linearly with the number of training examples. It is one of the fastest SVM solvers around for solving linear and multiclass L2 regularized SVMs.

This package contains the header files and static libraries.

SHOGUN - is a new machine learning toolbox with focus on large scale kernel methods and especially on Support Vector Machines (SVM) with focus to bioinformatics. It provides a generic SVM object interfacing to several different SVM implementations. Each of the SVMs can be combined with a variety of the many kernels implemented. It can deal with weighted linear combination of a number of sub-kernels, each of which not necessarily working on the same domain, where an optimal sub-kernel weighting can be learned using Multiple Kernel Learning. Apart from SVM 2-class classification and regression problems, a number of linear methods like Linear Discriminant Analysis (LDA), Linear Programming Machine (LPM), (Kernel) Perceptrons and also algorithms to train hidden markov models are implemented. The input feature-objects can be dense, sparse or strings and of type int/short/double/char and can be converted into different feature types. Chains of preprocessors (e.g. substracting the mean) can be attached to each feature object allowing for on-the-fly pre-processing.

SHOGUN comes in different flavours, a stand-a-lone version and also with interfaces to Matlab(tm), R, Octave, Readline and Python. This package includes the developer files required to create stand-a-lone executables.

LIBSVM, a machine-learning library, is an easy-to-use package for support vector classification, regression and one-class SVM. It supports multi-class classification, probability outputs, and parameter selection.

This package contains the development header files.

Torch is a machine-learning library, written in C++. Its aim is to provide the state-of-the-art of the best algorithms.

• Many gradient-based methods, including multi-layered perceptrons, radial basis functions, and mixtures of experts. Many small "modules" (Linear module, Tanh module, SoftMax module, ...) can be plugged together.
• Support Vector Machine, for classification and regression.
• Distribution package, includes Kmeans, Gaussian Mixture Models, Hidden Markov Models, and Bayes Classifier, and classes for speech recognition with embedded training.
• Ensemble models such as Bagging and Adaboost.
• Non-parametric models such as K-nearest-neighbors, Parzen Regression and Parzen Density Estimator.

This package is the Torch development package (header files and static library.)

Vision with Generic Algorithms (VIGRA) is a computer vision library that puts its main emphasis on flexible algorithms, because algorithms represent the principle know-how of this field. The library was consequently built using generic programming as introduced by Stepanov and Musser and exemplified in the C++ Standard Template Library. By writing a few adapters (image iterators and accessors) you can use VIGRA's algorithms on top of your data structures, within your environment.

This package contains the header and development files needed to build programs and packages using VIGRA.

This package provides graphical computation for Torch.

This package also ships a graphviz interface, you need not graphviz to be able to use this library but, if you have it, you will be able to display the graphs that you have created.

"image" is the Torch7 distribution package for processing images. It contains a wide variety of functions divided into the following categories:

• Saving and loading images as JPEG, PNG, PPM and PGM;
• Simple transformations like translation, scaling and rotation;
• Parameterized transformations like convolutions and warping;
• Simple Drawing Routines like drawing text or a rectangle on an image;
• Graphical user interfaces like display and window;
• Color Space Conversions from and to RGB, YUV, Lab, and HSL;
• Tensor Constructors for creating Lenna, Fabio and Gaussian and Laplacian kernels;

Note that unless specified otherwise, this package deals with images of size nChannel x height x width.

This package provides an easy and modular way to build and train simple or complex neural networks using Torch Framework:

• Modules are the bricks used to build neural networks. Each are themselves neural networks, but can be combined with other networks using containers to create complex neural networks:

• Module: abstract class inherited by all modules.

• Containers: container classes.
• Transfer functions: non-linear functions.
• Simple layers: simple network layer like Linear.
• Table layers: layers for manipulating tables.

• Convolution layers: several kinds of convolutions.

• Criterions compute a gradient according to a given loss function given an input and a target:

• Criterions: a list of all criterions.

• MSECriterion: the Mean Squared Error criterion used for regression;

• ClassNLLCriterion: the Negative Log Likelihood criterion used for classification.

• Additional documentation:

• Overview of the package essentials including modules, containers and training.

• Training: how to train a neural network using optim.
• Testing: how to test your modules.
• Experimental Modules: a package containing experimental modules and criteria.

This package is a core part of the Torch Framework.

This package provides graphical computation for nn library in Torch. The aim of this library is to provide users of nn package with tools to easily create complicated architectures. Any given nn module is going to be bundled into a graph node. The __call__ operator of an instance of nn.Module is used to create architectures as if one is writing function calls.

This package contains several optimization routines and a logger for Torch.

The following algorithms are provided:

• Stochastic Gradient Descent
• Averaged Stochastic Gradient Descent
• L-BFGS
• Congugate Gradients
• AdaDelta
• AdaGrad
• Adam
• AdaMax
• FISTA with backtracking line search
• Nesterov's Accelerated Gradient method
• RMSprop
• Rprop
• CMAES All these algorithms are designed to support batch optimization as well as stochastic optimization. It's up to the user to construct an objective function that represents the batch, mini-batch, or single sample on which to evaluate the objective.

This package provides also logging and live plotting capabilities via the optim.Logger() function. Live logging is essential to monitor the network accuracy and cost function during training and testing, for spotting under- and over-fitting, for early stopping or just for monitoring the health of the current optimisation task.

A pure Lua REPL (Read,Eval,Print-Loop) for LuaJIT, with heavy support for Torch types. It uses Readline for tab completion.

This package contains backend files to support the command line frontend 'th'.

Lua is pretty compact in terms of built-in functionalities: this package extends the table and string libraries, and provide other general purpose tools (progress bar, ...).

This package ships a set of useful extensions to Lua for Torch Framework.

Le paquet MCL implémente l'algorithme MCL (Markov Cluster Algorithm) et fournit les utilitaires nécessaires à la manipulation de matrices creuses (type de matrices essentielles dans l'algorithme MCL) et à la réalisation d’expérimentations dans les grappes.

MCL est utilisé dans divers domaines comme en biologie (détection de familles de protéines, génomiques), en informatique (grappe de nœuds dans un réseau pair à pair) ou la linguistique (analyse de textes).

Given an observed image containing a chessboard or a grid of circles, mrgingham locates the board in the image, and precisely computes the location of the chessboard corners (or circle centers). This is similar to the routines in OpenCV, but is faster and more robust.

This package provides the user-facing tools

Ce paquet fournit une fonction pour travailler avec des algorithmes génétiques dans Octave, un logiciel de calcul numérique. Il offre la fonction ga(), qui fonctionne de la même manière que d'autres fonctions d'optimisation dans Octave.

Ce paquet de greffon d’Octave fait partie du projet Octave-Forge.

Amp is an open-source package designed to easily bring machine-learning to atomistic calculations. This project is being developed at Brown University in the School of Engineering, primarily by Andrew Peterson and Alireza Khorshidi, and is released under the GNU General Public License. Amp allows for the modular representation of the potential energy surface, allowing the user to specify or create descriptor and regression methods.

Amp is designed to integrate closely with the Atomic Simulation Environment (ASE). As such, the interface is in pure python, although several compute-heavy parts of the underlying code also have fortran versions to accelerate the calculations. The close integration with ASE means that any calculator that works with ASE ─ including EMT, GPAW, DACAPO, VASP, NWChem, and Gaussian ─ can easily be used as the parent method.

This package provides the python 3 modules.

Fast Artificial Neural Network Library is a free open source neural network library, which implements multilayer artificial neural networks in C with support for both fully connected and sparsely connected networks.

This package contains the Python 3 bindings for FANN.

Python3-genetic provides genetic algorithms for Python3, as often used in artificial intelligence. It should be able to solve any problem that consists in minimizing functions.

You'll find some demos using Genetic in this package, including an impressively simple program that provides a solution to the well-known TSP (Travelling Salesman Problem). Also, make sure to read demo/genetic_demo_2.py for the list of the special "magic" genes that make Genetic really fun and ... living !

Keras est une bibliothèque de Python pour l’apprentissage automatique basée sur les réseaux de neurones (DNN) profonds (multicouches), qui adopte une conception minimaliste et modulaire avec un accent sur l’expérimentation rapide.

Les fonctionnalités des DNN, telles que les couches de neurones, les fonctions de coût, les optimiseurs, les schémas d’initialisation, les fonctions d’activation et les fonctions de régularisation sont disponibles dans Keras sous forme de modules autonomes qui peuvent être interconnectés à la demande pour créer des modèles de séquence ou des architectures plus complexes. Keras prend en charge les réseaux de neurones à convolution (CNN, utilisé pour la classification respective de reconnaissance d’image) et les réseaux de neurones récurrents (RNN, adapté à l’analyse de séquences comme dans le traitement du langage naturel).

Elle fonctionne comme une couche d’abstraction au-dessus de Theano (compilateur d’expressions mathématiques) par défaut, ce qui lui permet d’accélérer les calculs en utilisant des périphériques (GP)GPU. Autrement, Keras peut s’exécuter sur TensorFlow de Google (pas encore disponible dans Debian).

Lasagne is a Python library to build and train deep (multi-layered) artificial neural networks on the top of Theano (math expression compiler). In comparison to other abstraction layers for that like e.g. Keras, it abstracts Theano as little as possible.

Lasagne supports networks like Convolutional Neural Networks (CNN, mostly used for image recognition resp. classification) and the Long Short-Term Memory type (LSTM, a subtype of Recurrent Neural Networks, RNN).

This package contains the modules for Python 3.

Il s’agit d’un cadriciel en Python de traitement de données pour construire des logiciels complexes de traitement de données en combinant des algorithmes d’apprentissage automatique largement utilisés dans des tuyauteries et réseaux. Les algorithmes implémentés incluent l'analyse en composantes principales (PCA), l'analyse en composantes indépendantes (ICA), Slow Feature Analysis (SFA — analyse des variations lentes), Independent Slow Feature Analysis (ISFA), Growing Neural Gas (GNG — réseau neuronal artificiel incrémental), l’analyse factorielle, l’analyse discriminante linéaire de Fisher (FDA) et les classifieurs gaussiens.

mlpy provides high level procedures that support, with few lines of code, the design of rich Data Analysis Protocols (DAPs) for preprocessing, clustering, predictive classification and feature selection. Methods are available for feature weighting and ranking, data resampling, error evaluation and experiment landscaping.

mlpy includes: SVM (Support Vector Machine), KNN (K Nearest Neighbor), FDA, SRDA, PDA, DLDA (Fisher, Spectral Regression, Penalized, Diagonal Linear Discriminant Analysis) for classification and feature weighting, I-RELIEF, DWT and FSSun for feature weighting, RFE (Recursive Feature Elimination) and RFS (Recursive Forward Selection) for feature ranking, DWT, UWT, CWT (Discrete, Undecimated, Continuous Wavelet Transform), KNN imputing, DTW (Dynamic Time Warping), Hierarchical Clustering, k-medoids, Resampling Methods, Metric Functions, Canberra indicators.

Ce paquet fournit les liaisons Python⋅3 pour la bibliothèque OpenCV (Open Computer Vision).

La bibliothèque Open Computer Vision est un ensemble d'algorithmes et de codes d'exemple pour divers problèmes de vision par ordinateur. La bibliothèque est compatible avec IPL (Image Processing Library d'Intel) et, s'il est disponible, IPP d'Intel (Integrated Performance Primitives) pour de meilleures performances.

OpenCV fournit des types de données et des opérateurs portables de bas niveau et un ensemble de fonctionnalités de haut niveau pour l'acquisition vidéo, le traitement et l'analyse d'image, l'analyse structurale, l'analyse du mouvement et le suivi d'objet, la reconnaissance d’objet, la calibration de caméra et la reconstruction⋅3D.

scikit-learn is a collection of Python modules relevant to machine/statistical learning and data mining. Non-exhaustive list of included functionality:

• Gaussian Mixture Models
• Manifold learning
• kNN
• SVM (via LIBSVM)

This package contains the Python 3 version.

statsmodels Python3 module provides classes and functions for the estimation of several categories of statistical models. These currently include linear regression models, OLS, GLS, WLS and GLS with AR(p) errors, generalized linear models for several distribution families and M-estimators for robust linear models. An extensive list of result statistics are available for each estimation problem.

Thinc est une bibliothèque d’apprentissage automatique servant de dorsal pour spaCy https://spacy.io. Elle présente un modèle linéaire ayant passé l’épreuve du feu pour de larges problèmes d’apprentissage à vecteurs de support et un modèle de réseau neuronal encore en développement pour spaCy version 2.0 https://spacy.io/usage/v2.

Thinc est une boîte à outils pratique pour mettre en œuvre les modèles qui suivent l’architecture « Embed, encode, attend, predict » (intégrer, encoder, attendre, prédire). Elle est conçue pour être facile à installer, efficace du point de vue utilisation du CPU et optimisée pour le traitement automatique du langage naturel et l’apprentissage profond avec du texte, en particulier, pour une entrée structurée hiérarchiquement.

PyTorch is a Python package that provides two high-level features:

(1) Tensor computation (like NumPy) with strong GPU acceleration (2) Deep neural networks built on a tape-based autograd system

You can reuse your favorite Python packages such as NumPy, SciPy and Cython to extend PyTorch when needed.

This is the CPU-only version of PyTorch (Python interface).

This package consists of a small extension library of optimized sparse matrix operations with autograd support.

This package installs the library for Python 3.

Vision with Generic Algorithms (VIGRA) is a computer vision library that puts its main emphasis on flexible algorithms, because algorithms represent the principle know-how of this field. The library was consequently built using generic programming as introduced by Stepanov and Musser and exemplified in the C++ Standard Template Library. By writing a few adapters (image iterators and accessors) you can use VIGRA's algorithms on top of your data structures, within your environment.

This package exports the functionality of the VIGRA library to Python3.

Ce paquet a été créé pour publier l’algorithme de réseau neuronal robuste TAO pour les utilisateurs de R. Il s’est étoffé et peut être intéressant pour les utilisateurs qui veulent mettre en œuvre leurs propres algorithmes d’apprentissage ainsi que ceux dont les besoins se situent uniquement en « espace utilisateur ».

Le paquet bayesm couvre plusieurs modèles importants d’applications de commerce et de microéconométrie. Ce paquet comprend :

 – régression bayésienne (dep var univariés ou multivariés) ;
 – logit multinomial (MNL) et probit multinomial (MNP) ;
 – probit multivarié ;
 – normale multivariée de mélange ;
 – modèle linéaire hiérarchique avec probabilité normale et covariables ;
 – logit multinomial hiérarchique avec mélange de probabilités normales et
   de covariables ;
 – analyse bayésienne de données conjointes basées sur le choix ;
 – traitement bayésien de modèles linéaires de variables instrumentales ;
 – analyse de données d’enquête ordinales multivariées avec hétérogénéité
   d’utilisation d’échelle (comme dans Rossi et al., JASA (01)).

Pour plus de référence, consulter le livre des auteurs « Bayesian Statistics and Marketing » par Allenby, McCulloch et Rossi.

Le paquet class fournit des fonctions et des ensembles de données pour prendre en charge le douzième chapitre à propos de « classification » du livre « Modern Applied Statistics with S » (4ème édition) de W.N. Venables et B.D. Ripley. L’URL suivant fournit davantage de détails sur le livre : http://www.stats.ox.ac.uk/pub/MASS4

Ce paquet fournit les fonctions et les jeux de données pour l'analyse de grappe, écrit à l'origine par Peter Rousseeuw, Anja Struyf et Mia Hubert.

Ce paquet fait partie d'un ensemble de paquets recommandés par R et fournis avec les sources de R lui-même.

This package implements extensions to Freund and Schapire's AdaBoost algorithm and Friedman's gradient boosting machine. Includes regression methods for least squares, absolute loss, t-distribution loss, quantile regression, logistic, multinomial logistic, Poisson, Cox proportional hazards partial likelihood, AdaBoost exponential loss, Huberized hinge loss, and Learning to Rank measures (LambdaMart).

Le paquet MASS fournit des fonctions et ensembles de données pour prendre en charge le livre « Modern Applied Statistics with S » (quatrième édition) de W.N. Venables et B.D. Ripley. L’URL suivant fournit plus de détails à propos du livre : http://www.stats.ox.ac.uk/pub/MASS4

Il s’agit d’un ensemble de routines pour GNU R qui met en œuvre divers modèles statistiques et économétriques d’estimation de Monte-Carlo par chaînes de Markov (MCMC), qui permettent de « résoudre » des modèles qui seraient autrement insolubles avec les techniques traditionnelles, en particulier les problèmes dans les statistiques (où un ou plusieurs « a priori » sont utilisés comme partie de la procédure, au lieu de l’hypothèse d’ignorance du « vrai » point estimé), quoique que MCMC peut aussi être utilisé pour résoudre des problèmes de statistiques fréquentistes avec des a priori non instructifs. Les techniques de MCMC sont aussi préférables à l’estimation directe en présence de données manquantes.

Sont actuellement implémentées plusieurs routines d’inférence écologique (EI) (pour l’estimation des attributs ou comportement au niveau individuel à partir de données agrégées, tels que les rapports électoraux ou les résultats de recensement), ainsi que des modèles de panel linéaire classique et des données transversales, quelques routines de visualisation pour des diagnostics d’EI, deux modèles de théories de réponse à l’item (ou estimation par point optimal — ideal-point estimation), l’analyse de facteurs métriques, ordinaux et de réponse mixte, et des modèles de régression gaussienne (linéaire) et de Poisson, de régression logistique (ou logit) et des modèles probit de réponses binaires ou ordinales.

Les paquets suggérés (r-cran-bayesm, -eco et -mnp) fournissent des modèles supplémentaires pouvant être utiles aux utilisateurs de ce paquet.

An implementation of evaluation metrics in R that are commonly used in supervised machine learning. It implements metrics for regression, time series, binary classification, classification, and information retrieval problems. It has zero dependencies and a consistent, simple interface for all functions.

This GNU R package provices a collection of artificial and real-world machine learning benchmark problems, including, e.g., several data sets from the UCI repository.

Interface to a large number of classification and regression techniques, including machine-readable parameter descriptions. There is also an experimental extension for survival analysis, clustering and general, example-specific cost-sensitive learning. Generic resampling, including cross-validation, bootstrapping and subsampling. Hyperparameter tuning with modern optimization techniques, for single- and multi-objective problems. Filter and wrapper methods for feature selection. Extension of basic learners with additional operations common in machine learning, also allowing for easy nested resampling. Most operations can be parallelized.

MNP est un paquet de R qui adapte des modèles probit multinomiaux bayésiens (MNP) à l’aide de méthodes de Monte-Carlo par chaînes de Markov (MCMC). Selon le modèle probit multinomial standard, il peut aussi adapter des modèles avec différents ensembles de choix pour chaque observation et trier complètement ou partiellement toutes les alternatives disponibles. L'estimation se fonde sur l'algorithme d'augmentation marginale efficace des données développé par Imai et van Dyk (2004).

Fonctions pour adapter des modèles de Markov temporels cachés et continus à des données longitudinales. Les taux de transitions de Markov et le processus de sortie de Markov caché peuvent être modélisés en terme de co- variables. Une variété de schémas d'observation sont permis, qui comprennent une surveillance des processus à des temps arbitraires, une observation complète des processus, et des états censurés.

Bayesian nonstationary, semiparametric nonlinear regression and design by treed Gaussian processes (GPs) with jumps to the limiting linear model (LLM). Special cases also implemented include Bayesian linear models, CART, treed linear models, stationary separable and isotropic GPs, and GP single-index models. Provides 1-d and 2-d plotting functions (with projection and slice capabilities) and tree drawing, designed for visualization of tgp-class output. Sensitivity analysis and multi-resolution models are supported. Sequential experimental design and adaptive sampling functions are also provided, including ALM, ALC, and expected improvement. The latter supports derivative-free optimization of noisy black-box functions.

Torch is a scientific computing framework with wide support for machine learning algorithms that puts GPUs first. It is easy to use and efficient, thanks to an easy and fast scripting language, LuaJIT, and an underlying C/CUDA implementation.

A summary of core features:

• a powerful N-dimensional array
• lots of routines for indexing, slicing, transposing, ...
• amazing interface to C, via LuaJIT
• linear algebra routines
• neural network, and energy-based models
• numeric optimization routines
• Fast and efficient GPU support
• Embeddable, with ports to iOS, Android and FPGA backends

The goal of Torch is to have maximum flexibility and speed in building your scientific algorithms while making the process extremely simple. Torch comes with a large ecosystem of community-driven packages in machine learning, computer vision, signal processing, parallel processing, image, video, audio and networking among others, and builds on top of the Lua community.

At the heart of Torch are the popular neural network and optimization libraries which are simple to use, while having maximum flexibility in implementing complex neural network topologies. You can build arbitrary graphs of neural networks, and parallelize them over CPUs and GPUs in an efficient manner.

This package is a metapackage, which pulls the following core and free modules for you: cwrap, paths, sys, xlua, torch7, nn, graph, nngraph, optim, sundown, dok, trepl, image.

Note that cutorch (CUDA backend for torch) and cunn (CUDA backend for neural network) are not present in this metapacakge - they will be shipped in the torch-core-contrib metapackage in the future.

Toulbar2 is an exact discrete optimization tool for Graphical Models such as Cost Function Networks, Markov Random Fields, Weighted Constraint Satisfaction Problems and Bayesian Nets.

Weka is a collection of machine learning algorithms in Java that can either be used from the command-line, or called from your own Java code. Weka is also ideally suited for developing new machine learning schemes.

Implemented schemes cover decision tree inducers, rule learners, model tree generators, support vector machines, locally weighted regression, instance-based learning, bagging, boosting, and stacking. Also included are clustering methods, and an association rule learner. Apart from actual learning schemes, Weka also contains a large variety of tools that can be used for pre-processing datasets.

This package contains the binaries and examples.

Le kit d’échantillonnage adaptatif (ASK) est un ensemble d'outils pour échantillonner de grands espaces expérimentaux. Lorsque l'espace est petit, la réponse peut être mesurée pour chaque point de l'espace. Lorsque l'espace est grand, la réalisation d'une mesure exhaustive n'est pas possible soit en terme de temps d'exécution, soit de faisabilité. ASK essaye de trouver de bonnes approximations de la réponse en échantillonnant seulement une petite portion de l'espace. ASK dispose de multiples algorithmes d'apprentissage actif pour prioriser l'exploration des parties intéressantes de l'espace expérimental.

Dlib is a general purpose cross-platform open source software library written in the C++ programming language. It now contains software components for dealing with networking, threads, graphical interfaces, complex data structures, linear algebra, statistical machine learning, image processing, data mining, XML and text parsing, numerical optimization, Bayesian networks, and numerous other tasks.

This package contains the development headers.

DLPack is an open in-memory tensor structure to for sharing tensor among frameworks. DLPack enables

• Easier sharing of operators between deep learning frameworks.
• Easier wrapping of vendor level operator implementations, allowing collaboration when introducing new devices/ops.
• Quick swapping of backend implementations, like different version of BLAS
• For final users, this could bring more operators, and possibility of mixing usage between frameworks.

DLPack do not intend to implement of Tensor and Ops, but instead use this as common bridge to reuse tensor and ops across frameworks.

Use FannJ if you have an existing ANN from the FANN project (libfann2) that you would like to access from Java. There are several GUI tools that will help you create and train an ANN.

fclib is an open source collection of Frictional Contact (FC) problems stored in a specific HDF5 format, and an open source light implementation of Input/Output functions in C Language to read and write problems.

The goal of this work is to set up a collection of 2D and 3D Frictional Contact (FC) problems in order to set up a list of benchmarks; provide a standard framework for testing available and new algorithms; and share common formulations of problems in order to exchange data.

Fclib is an open-source scientific software primarily targeted at modeling and simulating nonsmooth dynamical systems

This package includes the libfclib development headers.

Intel(R) Math Kernel Library for Deep Neural Networks (Intel(R) MKL-DNN) is an open source performance library for deep learning applications. The library accelerates deep learning applications and framework on Intel(R) architecture. Intel(R) MKL-DNN contains vectorized and threaded building blocks which you can use to implement deep neural networks (DNN) with C and C++ interfaces.

DNN functionality optimized for Intel architecture is also included in Intel(R) Math Kernel Library (Intel(R) MKL). API in this implementation is not compatible with Intel MKL-DNN and does not include certain new and experimental features.

One can choose to build Intel MKL-DNN without binary dependency. The resulting version will be fully functional, however performance of certain convolution shapes and sizes and inner product relying on SGEMM function may be suboptimal.

This package contains the header files, and symbol links to the shared object.

Given an observed image containing a chessboard or a grid of circles, mrgingham locates the board in the image, and precisely computes the location of the chessboard corners (or circle centers). This is similar to the routines in OpenCV, but is faster and more robust.

This package provides the development C++ libraries

XGBoost is an optimized distributed gradient boosting library designed to be highly efficient, flexible and portable. It implements machine learning algorithms under the Gradient Boosting framework. XGBoost provides a parallel tree boosting (also known as GBDT, GBM) that solve many data science problems in a fast and accurate way. The same code runs on major distributed environment (Kubernetes, Hadoop, SGE, MPI, Dask) and can solve problems beyond billions of examples.

LIBXSMM is a library targeting Intel Architecture for specialized dense and sparse matrix operations, and deep learning primitives.

This package contains the tools, static libraries and header files.

Various definitions for a high-dimensional median exist and this Python package provides a number of fast implementations of these definitions. Medians are extremely useful due to their high breakdown point (up to 50% contamination) and have a number of nice applications in machine learning, computer vision, and high-dimensional statistics.

This package currently has implementations of medoid and geometric median with support for missing data using NaN.

imbalanced-learn is a Python package offering a number of re-sampling techniques commonly used in datasets showing strong between-class imbalance.

It is compatible with scikit-learn and is part of scikit-learn-contrib projects.

The liac-arff module implements functions to read and write ARFF files in Python. It was created in the Connectionist Artificial Intelligence Laboratory (LIAC), which takes place at the Federal University of Rio Grande do Sul (UFRGS), in Brazil.

ARFF (Attribute-Relation File Format) is an file format specially created for describing datasets which are used commonly for machine learning experiments and software. This file format was created to be used in WEKA, the best representative software for machine learning automated experiments.

Given an observed image containing a chessboard or a grid of circles, mrgingham locates the board in the image, and precisely computes the location of the chessboard corners (or circle centers). This is similar to the routines in OpenCV, but is faster and more robust.

This package provides the Python interfaces

Ce métapaquet installe les paquets de Debian Science concernant le calcul numérique. Les paquets fournissent un système de calcul et visualisation orienté matrice pour le calcul scientifique et l’analyse de données. Ces paquets sont similaires aux systèmes commerciaux tels que Matlab et IDL.

Ce paquet fait partie du mélange exclusif « Debian Science » et installe les paquets concernant les statistiques. Cette tâche est une tâche générale pouvant être utile pour n’importe quels travaux scientifiques. Elle dépend de beaucoup de paquets R ainsi que d’autres outils utiles pour faire des statistiques. De plus la tâche « Science Mathematics » est suggérée pour installer facultativement tous les logiciels relatifs aux mathématiques.

Ce métapaquet installe les paquets de Debian Science concernant la composition typographique. L’utilisateur peut aussi être intéressé par field::physics de debtag.

Caffe is a deep learning framework made with expression, speed, and modularity in mind. It is developed by the Berkeley AI Research Lab (BAIR) and community contributors.

This metapackage pulls CUDA version of caffe:

• caffe-tools-cuda
• libcaffe-cuda*

• python3-caffe-cuda And suggests these packages:

• libcaffe-cuda-dev

• caffe-doc

Note, this CUDA version cannot co-exist with the CPU_ONLY version.

spaCy is a library for advanced Natural Language Processing in Python and Cython. It’s built on the very latest research, and was designed from day one to be used in real products. spaCy comes with pre-trained statistical models and word vectors, and currently supports tokenization for 30+ languages. It features the fastest syntactic parser in the world, convolutional neural network models for tagging, parsing and named entity recognition and easy deep learning integration.

Streamlit lets you create apps for your machine learning projects with deceptively simple Python scripts. It supports hot-reloading, so your app updates live as you edit and save your file. No need to mess with HTTP requests, HTML, JavaScript, etc. All you need is your favorite editor and a browser.

Orange is a component-based data mining software. It includes a range of data visualization, exploration, preprocessing and modeling techniques. It can be used through a nice and intuitive user interface or, for more advanced users, as a module for Python programming language.

FLANN is a library for performing fast approximate nearest neighbor searches in high dimensional spaces. It contains a collection of algorithms we found to work best for nearest neighbor search and a system for automatically choosing the best algorithm and optimum parameters depending on the dataset.

PyBrain is a modular machine learning library for Python. Its goal is to offer flexible, easy-to-use yet still powerful algorithms for Machine Learning Tasks and a variety of predefined environments to test and compare your algorithms.

PyBrain currently features algorithms for Supervised Learning, Unsupervised Learning, Reinforcment Learning and Black-box Optimization.