Arctic Analyses 2014 - Heading for the Future!
Program, Ilulissat, March 2014
To see the actual talks, head over to the password protected version (only for participants) - enter here
Session 1: Big data
Vagelis Papalexakis: Big Data Tutorial, I-II
Morten Rasmussen: We love to take pictures of our kids; how can we use photos to improve their medical care?
Giorgio Tomasi: How do we incorporate chemical knowledge to separate good data from bad as a preprocessing step?
Kristian Liland: Big Data is already knocking on the door of chemometrics
Mary Beth Seasholz: Videos – available on request
Session 2: Modeling strategies
Tormod Næs: A discussion of different modelling paradigms and strategies in statistics and chemometrics
Marta Bevilacqua: Local Modeling in dealing with classification issues
Marina Cocchi: Class-modelling vs discriminant classification
Ulf Indahl: Seeking subspaces suitable for predictive modeling by the ‘classical’ approaches to regression and classification
Session 3: Dealing with Dirty Data
Alberto Ferrer: Dealing with Dirty Data Session: Introduction and motivation / Grey Modelling and Dirty Data
Ellen F Mosleth: A cycling process as a scientific strategy
Onno de Noord: Representativity issues in process chemometrics.
Federico Marini: It’s a dirty job, but somebody’s gotta do it
Åsmund Rinnan: How to make the perfect calibration the first time?
Ingunn Berget: Cluster analysis as a tool for dirty data
Cyril Ruckebusch: Single-molecule fluorescence imaging data: can we help?
Session 4: Data Fusion
Age Smilde: How and when to link data sets new possibilities for linking data, and new possibilities for modelling modes. When do we need to fuse data?
Evrim Acar: Challenges in Data fusion based on Coupled Matrix and Tensor Factorization Shared and unshared factors, when to fuse data, weighing of data sets
Ingrid Måge: Challenges in modelling and interpretation of multiblock data
Common and unique factors, interactions between data blocks, interpretation of complex models
Alessandra Biancolillo: Multiblock classification Combining multiblock models and LDA, evaluation of block contributions, plots/graphics for interpretation
Johan Westerhuis: Variable selection in multi set data. How to cope with different measurement errors, use of resampling techniques
Thomas Skov: Smart pre-processing and variable selection. Using the additional information in e.g. control samples to preprocess data and select variables in a clever way.
The Secret Session – the Future of Science
Henk Kiers & Age Smilde
Code of conduct
To encourage open communication, each participant agrees that any information presented is a private communication from the individual making the contribution and is presented with the restriction that such information is not for public use. If interesting ideas emerge during the conference, all the main contributors should be acknowledged, contacted and invited for further development of those ideas.