Welcome to the Amyloid workshop: User-friendly analysis of spectroscopy data with Quasar - multivariate statistics and machine learning.
LINXS, in collaboration with the SMIS beamline at SOLEIL and the Biolab from the University of Ljubljana, is organising a 3-half day hands-on workshop to introduce the QUASAR software, to address the infrared user community's need for a user-friendly and open-source software for data analysis
This 3-half day hands-on training will be fully digital during January 13-15, 2021. Participants will be selected on a fair distribution basis across the research groups and giving priority to early career.
During this workshop, we will focus on spectroscopic data analysis. The workshop is targeted at hyper-spectral imaging users (current, future or potential) working on biomedical applications, material-engineering, physical-chemical sciences, and more. The purpose of this training is to provide a practical introduction to the QUASAR software, using tutorials and examples on synchrotron data sets as well as real-life Raman and IR imaging datasets.
QUASAR is an open-source software for hyper-spectral imaging techniques (based on the Orange machine learning and data visualization suite).
QUASAR allows user-friendly analysis using visual programming. Routine tools like baseline correction, normalization, different versions of EMSC, differentiation and smoothing can be combined with multivariate statistical and machine learning methods, such as principal component analysis or various clustering methods. Savable and shareable workflows ensure consistent analysis across different projects, or the development of different analysis to same large dataset. Visualization tools enable quick inspection of the data and the results of the analyses.
The goal of the workshop is to teach the basic operations of chemical imaging to prepare the student to generate and interpret such images using QUASAR, new free software.
The workshop will bring together curious students and young researchers and introduce them to essential data mining and machine learning concepts in spectroscopic data analysis. Participant will learn about data visualization and machine learning with Quasar. Upon completion, participants will be able to analyze your own data and use them to develop predictive models. The workshop will be hands-on, with examples or own data
·Data exploration and visualization.
·Clustering, uncovering of groups in data.
·Classification and predictive modeling.
·3-day theory/hands-on course on key approaches of data science
·Free software and data sets used during the course
·Certificate of attendance
Applications are now open, and will be closed 10th of January 2021.
The workshop is limited to max 20 people selected on a fair distribution basis across the research groups, giving priority to early career scientists.
There is no fee for digital participation.
Dr. Ferenc Borondics, SOLEIL, France
Dr. Christophe Sandt, SOLEIL, France
Dr. Marko Toplak, University of Ljubljana, Slovenia
The Amyloid working group at LINXS and AI Lund (website)
Day 1, January 13, 2- 6 pm
Getting started with Quasar (installation, basic Orange and Quasar functionality)
Spectral Preprocessing
Advanced visualization
Statistical inspection of data
PCA, PCA imaging
Hands-on work with participants' data
Day 2, January 14, 2- 6 pm
Supervised learning
Introduction to supervised learning
Classification of spectra and hyperspectral datasets using various methods
Model inspection and cross-validation
Common errors
Hands-on work with participants' data
Day 3, January 15, 2- 6 pm
Unsupervised learning
Introduction to unsupervised learning
Clustering of spectra and hyperspectral datasets using various methods
Common errors
Partial Least Squares regressionxx
Hands-on work with participants' data