machine learning framework

Category Intelligent Software>Fuzzy Logic Systems/Tools and Intelligent Software>Data Mining Systems/Tools

Abstract machine learning framework (mlf) is a universal tool for creating understandable computational models from data.

It combines fuzzy logic based machine learning methods. Fully implemented in C++, mlf is integrated into Mathematica (Note 1).

This product can be used by professionals, who want to extract models from data:

1) Business and financial engineers.

2) Process and manufacturing engineers.

3) Quality assurance (QA) professionals.

It can also be used by Experts, who want to integrate data mining and machine learning solutions, such as, knowledge engineers and machine learning experts.

Products differential advantage --

1) Product creates understandable, numerically optimized computational models.

2) Fuzzy logic based approaches create linguistic interpretable models, which can be computed like black boxes.

3) Mathematica’s unique computing and programming environment allows the quickest customization and configuration of the solutions integration.

Products features/capabilities include:

Parallel Data Mining --

1) Perform time-consuming machine learning tasks within mlf in parallel; without making the handling complex.

2) Explore different models in parallel.

3) Use several CPUs to assess the quality of a particular model.

Time-Series --

1) Build models on time series data using all available algorithms.

Kernel Methods --

1) Gaussian process regression: build non-linear regression models in the kernel way.

2) Support vector machines: build classification models in the kernel way.

Supervised Analysis --

1) Ridge Regression: Regression with built-in feature selection.

2) Additive regression and Boosting: refining models incrementally.

3) Quadratic regression models: build non-linear models which select the right features.

4) Neural networks: build highly accurate models for prediction tasks.

Unsupervised Analysis --

1) Self-organizing maps (SOMs): create two-dimensional plots of high dimensional data sets, pre-process large and noisy data sets, recall (one or more) missing values in the data.

2) Fuzzy C-means: creates a fuzzy segmentation of the data.

3) Ward clustering: a crisp, agglomerative clustering method.

4) Learning vector quantization.

Fuzzy logic --

1) Fuzzy decision trees: FS-ID3, a fuzzy variant of the ID3 (Iterative Dichotomiser 3) learning algorithm used to create decision trees.

2) Fuzzy rule generation: FS-FOIL, a fuzzy variant of Quinlan's FOIL (inductive logic programming) method.

3) Cluster descriptions: FS-MINER, a proprietary method from SCCH GmbH to find cluster descriptions.

4) Optimization of fuzzy controllers: RENO, a proprietary method from SCCH GmbH, which uses numerical optimization to find computationally accurate and robust fuzzy rules.

5) Different types of fuzzy sets, t-norms and inference (Mamdani, Sugeno, Tagaki-Sugeno-Kang).

Efficiently build your models --

1) Advanced functions for routine tasks.

2) Automated model testing.

3) Advanced data visualization.

4) ODBC Data import.

New features in mlf version 2.0 --

1) Mathematica 7 is required (Mathematica 8 compatible).

2) mlf was enhanced with new functions which allow parallel data mining tasks: cross-validation and the processing of the “Model Explorer”.

The Model Explorer allows you to automatically run a lot of time-consuming tasks “overnight” and afterwards look at the performance of the different models that were created and tested.

3) The range of available methods/algorithms was enlarged by kernel methods for classification and regression: support vector machines (SVMs) and Gaussian process regression.

4) A feature was added that allows the integration of user defined algorithms and performance measures into mlf.

5) New meta-Learning algorithms for feature selection and boosting algorithms are also available.

6) Furthermore, building models for time series data is more directly supported.

Note: Mathematica is a computer program used mainly in scientific, engineering and mathematical fields. It was originally conceived by Stephen Wolfram and developed by a team of mathematicians and programmers that he assembled and led. It is sold by Wolfram Research and its distributors.

Mathematica provides cross-platform support for tasks such as symbolic or numerical calculations, arbitrary precision arithmetic, data processing, and plotting. Mathematica offers a programming language which supports functional and procedural programming styles.

System Requirements

Contact manufacturer.


Manufacturer Web Site machine learning framework

Price Contact manufacturer.

G6G Abstract Number 20165R

G6G Manufacturer Number 102847