Data Science or not Data Science?
Welcome to LIPN Wiki on Big Data and Cloud Computing
With more and more data produced every day, we need to pay a special attention on the technologies to use in order to be able to analyze large amount of data. Big Data is often characterized by the 4 V for Volume, Variety, Velocity, Veracity that constitute challenges for the required tools.
Machine learning is to extract knowledge from data. In short it's a family of algorithms that transform data into model or description with the aim to predict or categorize data. In this field we use also analytics tools consisting to present informations in a more readable way as for the Square Predict project. Others projects related to big data and cloud computing are the Wendelin and the Resilience projects. This research has been supported by the French Foundation FSN, PIA Grant Big data-Investissements d'Avenir.
The wiki is related to our experience with the Grid5000, Teralab and CIRRUS testbeds for the study of the Software, Platform, Infrastructure and Network layers that push forward the Data Science field according to an experimental scientific method. We must not confuse the 'scientific problem' term and the 'System scientific problem' term to serve the scientific problem.
General discussion on Systems for Big-Data
Infrastructure, programming models, frameworks Tutorial on moving data around Grid'5000 Clustering geolocated data using Spark and DBSCAN (illustrative example) Spark and Scikit-Learn (illustrative example)
Our experience is with the following tools:
Spark-notebook Apache Spark Apache Flink TensorFlow Wendelin See also this link SlapOS IBM Bluemix
Testbeds we use in conjunction with our experimental method:
Grid5000 CIRRUS @ Université Sorbonne Paris Cité Teralab Amazon
Apache Spark : From models to platform
NEW: More interesting for you : Spark-clustering-notebook [1] Some Apache Spark implementations (since 2011/2012) How to use Spark on Grid5000 How to use Spark on Magi([2]) Spark on your own machine
SlapOS cloud
General information on SlapOS BOINC as a Service for the SlapOS Cloud: Tools and Methods Déploiement de la plate-forme SlapOS dans l'environnement Grid'5000 A Cloud-Hosted SaaS with SlapOS for BLAST Benchmark on Grid5000 See also this link and this journal paper Volunteer Cloud for e-Science Synthesis of the LIPN work for the FUI Resilience project
TeraLab
General information on TeraLab How to use TeraLab TeraLab and SlapOS and VFIB fees for managing your infrasructure.
Thesis corner
Leila Abidi: Revisiter les grilles de PCs avec des technologies du Web et le cloud computing (2015) [Walid Saad]: Gestion de données pour le calcul scientifique dans les environnements grilles et cloud (2016) [Tugdual Sarazin]: Apprentissage massivement distribué dans un environnement "Big data" (2016) [Mohammed Ghesmoune]: Fouille de flux de données massives. Application aux “BigData” d’assurance (2016) [Hippolyte Léger]: Apprentissage Relationnel Massif (2018)
