Open Source Java Machine Learning Software - Page 3

The KReator project is a collection of software systems, tools, algorithms and data structures for logic-based knowledge representation. Currently, it includes the software systems KReator and MECore and the library Log4KR: - KReator is an integrated development environment (IDE) for relational probabilistic knowledge representation languages such as Bayesian Logic Programs (BLPs), Markov Logic Networks (MLNs), Relational Maximum Entropy (RME), First-Order Probabilistic Conditional Logic (FO-PCL), and others. - MECore is a shell-based system that allows the user to create propositional knowledge bases, to perform a variety of belief change operations, and to query a knowledge base with respect to the principle of optimum entropy. - Log4KR is a library providing data structures to represent knowledge bases in various logic formalisms (propositional, relational, conditional, probabilistic, ...) and providing algorithms to perform reasoning operations

MultiViL is a tool for multi-view learning. It supports four classifiers (KNN, Naive-Bayes, Rochio and SVM-Perf), four view combining methods (Majority Voting, Borda Count, Dempster-Shafer theory of evidence and PSO) and provides many analisys tools.

P53 Cancer Rescue Project, University of California, Irvine , Samuel A. Danziger, Christopher Wassman, Faezeh Salehi Amiri, Roberta Baronio, Linda Hall, Rainer K. Brachmann, G. Wesley Hatfield, Peter Kaiser, Richard H. Lathrop

The Information Extraction Plugin allows the use of information extraction techniques within RapidMiner. It can be seen as an interface between natural language and IE- or datamining-methods, by extracting interesting information out of documents.

This software uses a genetic algorithm to "evolve" reactor designs for Schema's space-simulation game, Star-Made (http://star-made.org/). One of the more unique aspects of the game is that ship power management is not as simple as filling a cube with power generation blocks. This mechanism leads to difficulties in getting maximal power generation out of such reactors. This program rectifies this issue by using a self-organizing variant of brute force search. Much of the search space is pruned early on by a user-selectable fitness function. Within a few epochs, reactor output quickly converges to several sub-optimal, yet high-output reactors. Given enough time, the idea is that the optimal reactor configuration will be yielded.

REPOSITORY MOVED TO GITHUB: https://github.com/happyjack27/SwarmWars video sample: http://youtu.be/s5mLNbdBQGY A game where you evolve & compete AI swarms. The organisms use swarm intelligence & ant colony optimization. The organisms can communicate through 3-color signaling as well as by laying beacons. They can attack and repair other organisms. They can select mates, and they can gather and distribute food and material. This behavior is controlled by a genetically evolved neural net augmented with online back propagation learning. The back propagation learning uses a reward vector and plasticity matrix that is evolved as part of the genome. Long story short, the AI is pretty frickin' sophisticated. Players can take control of organisms, trade resources and organisms in a market, and aid evolution by selective breeding.

TreeLiker is a collection of fast algorithms for working with complex structured data in relational form. The data can, for example, describe large organic molecules such as proteins or groups of individuals such as social networks or predator-prey networks etc. The algorithms included in TreeLiker are unique in that, in principle, they are able to search given sets of relational patterns exhaustively, thus guaranteeing that if some good pattern capturing an important feature of the problem exists, it will be found. In experiments with real-life data, the algorithms were shown to be able to construct complete non-redundant sets of patterns for chemical datasets involving several thousands of molecules as well as for comparably large datasets from genomics or proteomics. The included relational learning algorithms are tailored towards so-called tree-like features for which some otherwise very hard sub-problems (NP-hard) become tractable.

TrimGA is a lightweight genetic algorithm library written in pure Java 6.0 that can be quickly applied to most optimization problems.

We moved to GitHub: https://github.com/grobid/grobid

libVMR is a class library written in Java which implements code generator for group method of data handling - GMDH. The library is intended for users, with machine learning skills. libVMR provides an effective framework for the research and development of data mining and predictive analytics. libVMR is based on the most popular neural network model with a higher generalization ability from kernel tricks - vector machine by Reshetov (VMR). The library has been designed to learn from data sets. Typical applications here are pattern recognition ( binary classification).

This Java software implements Profile Hidden Markov Models (PHMMs) for protein classification for the WEKA workbench. Standard PHMMs and newly introduced binary PHMMs are used. In addition the software allows propositionalisation of PHMMs.

This project consists in a set of challenges to recognize images acquired from 3d Lasers.

The recognition of Activities of Daily Living (ADL) has represented one of the most developed research areas in recent years. Its objective is to determine what daily activity is developed by the inhabitants of a smart environment. In this project, an ontology-based framework for the mining of ADL with a generic ontology and a modular architecture is proposed. The framework includes applications to: i) load multiple datasets available in literature, ii) to provide different methods for the segmentation of the activities, and iii) to transform the datasets into different ontological models.

Automatic Intelligent Learning Code.

Just carved a bunch of bytes and have no idea what they could be? Maybe ANNFiD can help. ANNFiD uses neural network to identify byte patterns. It can be trained and has a GUI to help in the process. The tool is still on a very early stage, but could improve exponentially with the help of the developer community

Java library devoted to handle Genetic Algorithms and Classifier Systems. It has been engineered to be used into agent based simulation models and to search bounded optimal solutions in wide solution spaces. It runs on distributed clusters.

The AdPreqFr4SL learning framework for Bayesian Network Classifiers is designed to handle the cost / performance trade-off and cope with concept drift. Our strategy for incorporating new data is based on bias management and gradual adaptation. Starting with the simple Naive Bayes, we scale up the complexity by gradually updating attributes and structure. Since updating the structure is a costly task, we use new data to primarily adapt the parameters and only if this is really necessary, do we adapt the structure. The method for handling concept drift is based on the Shewhart P-Chart. Project homepage: http://adpreqfr4sl.sourceforge.net

Alink is Alibaba’s scalable machine learning algorithm platform built on Apache Flink, designed for batch and stream data processing. It provides a wide variety of ready-to-use ML algorithms for tasks like classification, regression, clustering, recommendation, and more. Written in Java and Scala, Alink is suitable for enterprise-grade big data applications where performance and scalability are crucial. It supports model training, evaluation, and deployment in real-time environments and integrates seamlessly into Alibaba’s cloud ecosystem.

Angel is a high-performance distributed machine learning and graph computing platform based on the philosophy of Parameter Server. It is tuned for performance with big data from Tencent and has a wide range of applicability and stability, demonstrating an increasing advantage in handling higher-dimension models. Angel is jointly developed by Tencent and Peking University, taking account of both high availability in industry and innovation in academia. With a model-centered core design concept, Angel partitions the parameters of complex models into multiple parameter-server nodes and implements a variety of machine learning algorithms and graph algorithms using efficient model-updating interfaces and functions, as well as a flexible consistency model for synchronization. Angel is developed with Java and Scala. It supports running on Yarn. With PS Service abstraction, it supports Spark on Angel.

Using Machine Learning, AraRooter finds the three-lettered root of any Arabic lemma with around 84% accuracy.

This project aimed at creating framework and binary data format for etymological Arabic system. and will not continue hosted at sourceforge because the term of use determine me as enemy, so I am prohibited from using sourceforge services.

Highly reusable and extensible Decision-Tree (Max-Gain) framework comprising of comprehensive input-processing and display functionality. Handles nominal, linear, continuous data. For preliminary description, refer - http://sushain.com/blog/archives/