Projects

UnBBayes

UnBBayes is a probabilistic network framework written in Java. It has both a GUI and an API with inference, sampling, learning and evaluation. It supports BN, ID, MSBN, OOBN, HBN, MEBN/PR-OWL, structure, parameter and incremental learning.

UncertWeb

claimed by 52°North Initiative for Geospatial Open Source Software

UncertWeb is an EC funded research project running from Feb 2010 – Jan 2013 developing the uncertainty enabled model web. The model Web concept, formulated within the Global Earth Observation System of Systems activity envisages the integration of complex resources, such as data and models, to ... [More] construct complex models, composed of chains of model and data components exposed as web services. This offers exciting opportunities for model development in a more loosely coupled, component oriented manner, encouraging sharing, re-use and easy access. [Less]

GeoAR

claimed by 52°North Initiative for Geospatial Open Source Software

GeoAR is an (augmented reality) browser for geospatial data for the Android operating system. Its first version is based on a thesis in the context of OpenNoiseMap, which implemented clustering and visualisation of in-situ noise measurements using OpenGL ES, but that was extensively re-implemeneted and a lot more functionality added.

Greenland

claimed by 52°North Initiative for Geospatial Open Source Software

Greenland is a client-side tool to render quality-aware geospatial data, implemented as OpenLayers-based webpage. Check out the latest version at geoviqua.dev.52north.org/greenland/

codbrp

1) Summarise current cod reference points e.g. use and basis of their derivation 2) Identify inconsistencies, e.g. between i. biomass and fishing mortality limits and targets, ii. precautionary and limits reference points 3) Note lack of targets and suggest appropriate reference points 4) ... [More] Compare the reference points estimated for all cod stocks with the current ICES stocks. 5) Compare risk levels and uncertainty (e.g. for data-poor v. data rich stocks) 6) Evaluate robustness to climate change 7) Propose a consistent methodology to derive robust reference points based on SR Meta-analysis and life history characteristics [Less]

blipinv

A set of code accompanying WRR paper (Cardiff and Kitanidis 2009-in review) for estimation of parameter fields using a level-set framework. The code is in a working development state and has been tested on a few example problems prior to publication in WRR. We are always looking for people to ... [More] help test the code on further problems, though, as well as to help improve our algorithms! If you have comments, questions, suggestions, or would like to help with improvements to our numerical algorithms, please e-mail mcardiff@stanford.edu and Michael.Cardiff@gmail.com. The code is freely available to anyone interested in its use, however we request that anyone who distributes or utilizes the code references Cardiff and Kitanidis' (2009-in review) WRR paper in their work. [Less]

robly

Il progetto fornisce una funzione di MATLAB di facile utilizzo per risolvere problemi di ottimizzazione con le funzioni quadratiche omogenee di Lyapunov. Queste sono usate per lo studio della stabilità robusta per sistemi con struttura incerta. Per fare ciò, la funzione che abbiamo sviluppato usa: ... [More] YALMIP: A Toolbox for Modeling and Optimization for feasibility problem in MATLAB. J. Löfberg. In Proceedings of the CACSD Conference, Taipei, Taiwan, 2004. (Link) SEDUMI: Is a software package to solve optimization problems over symmetric cones. This includes linear, quadratic, second order conic and semidefinite optimization, and any combination of these. (Link) Per usare la nostra funzione è necessario scaricare ed installare i sopracitati due tools che si trovano ai seguenti indirizzi: Download Yalmip Download SeDuMi --- The project provide an easy-to-use MATLAB function to solve feasibility problems with quadratic homogeneous Lyapunov functions. These are used to study the robust stability for system with structured uncertainties. To accomplish this, the function we developed uses: YALMIP: A Toolbox for Modeling and Optimization for feasibility problem in MATLAB. J. Löfberg. In Proceedings of the CACSD Conference, Taipei, Taiwan, 2004. (Link) SEDUMI: Is a software package to solve optimization problems over symmetric cones. This includes linear, quadratic, second order conic and semidefinite optimization, and any combination of these. (Link) To use our function you need to download and install the two tools written above downloading them at the following URL: Download Yalmip Download SeDuMi [Less]

scalby

Scala Implementation of Bayesian Neworks (WorkInProgress) Implemented so far : - d-separation algorithm (with test cases) - Inference Algo - Pearl's Message passing algorithm for a Bayesian Network whose DAG is a tree.(with test cases) - Inference Algo - Pearl's Message passing ... [More] algorithm for a Bayesian Network whose DAG is Singly Connected.(with test cases) [Less]

libpgrl

LibPGThe PG library was intended to be a high-performance policy-gradient reinforcement learning library. Since the first version it has been extended to a number of value based RL algorithms, so the name is only historical. It is now a general RL library which implements, for example, natural actor ... [More] critic, and least squares policy iteration. It has been designed with large distributed RL systems in mind. It's not perfect, but it is pretty fast. API documentation and examples are provided. What libpg does NOT provide is model based planning algorithms such as value iteration, or real-time dynamic programming, or exact policy gradient. There is limited support for belief state tracking in the simulators/Cassandra/ directory (named because we use the POMDP file format created by Anthony Cassandra). One day I'd like to extend it to these situations, but that will require some uptake of the library. Project goalsProvide easy to use implementations of state-of-the-art RL algorithms for the non RL savvy, allowing immediate application to difficult industry problems. High performance, especially on multi-agent RL problems Extensible plug'n'play algorithms for research purposes Main algorithms implementedRL algorithms: SARSA QLearning Vanilla Policy-Gradient (online or batch, including line search) Natural actor-critic Least Squares TD-Q(\lambda) (for LSPI) Least Squares Policy Iteration Misc supporting algorithms: Line searches for batch mode Tikhonov Regularisation HMM based POMDP state estimation Finite history transformation [Less]