GeoDSc SPEC TRACK

//GeoDSc SPEC TRACK
GeoDSc SPEC TRACK2018-12-11T13:38:07+00:00

Specialization Track GeoDSc - GeoData Science

University of South Brittanny, Département Informatique

MANDATORY MODULES

Semesters 3 + 4

Specialization Track – GeoData Science –  with 24 ECTS is completed during semester 3, typically leading to a master’s thesis in line with the track and co-supervised at UBS together with PLUS. Semester 4 is dedicated to the Master thesis development and examination.

Source: https://commons.wikimedia.org/wiki/File:Deep_learning.png

Source: Wikimedia

Learning Outcomes: Upon completion of the module, students are able to:

  • understand the different machine learning problems and methods;
  • design for a given data analytics problem the appropriate solution to be used;
  • implement deep learning models within a standard framework.

Module content:

  • Machine Learning:
    • principles of supervised learning and other machine learning paradigms;
    • classification and regression, with discriminative and generative models;
    • dimension reduction and feature selection;
    • anomaly detection;
    • training strategies and evaluation protocols;
    • use of software libraries.
  • Deep Learning:
    • principles of neural networks;
    • optimization, regularization, and transfer;
    • main architectures (CNN, RNN, AE, GAN);
    • use of deep learning software frameworks.

Type of exam:  Assessment of individual lab assignment plus overview test.

The courses are taught as a combination of lectures with practical lab components.

Source: UBS-IRISA

Learning Outcomes: Upon completion of the module, students are able to:

  • understand advanced models and techniques for image processing;
  • solve realistic problems in computer vision.

Module content:

  • Image Processing:
    • principles of image processing and review of basic methods;
    • advanced structured representations (including graphs and trees);
    • efficient processing methods (non-local, multiscale).
  • Image Analysis:
    • global and local image features;
    • segmentation and object detection;
    • indexing and retrieval;
    • deep learning for computer vision.

Type of exam:  Assessment of individual lab assignment plus overview test.

The courses are taught as a combination of lectures with practical lab components.

Source: https://fr.wikipedia.org/wiki/Fichier:DARPA_Big_Data.jpg

Source: Wikipedia

Learning Outcomes: Upon completion of the module, students are able to:

  • understand the principles of knowledge discovery and the methods for data mining;
  • use software framework to design, implement and deploy a solution for big data analytics.

Module content:

  • Data Mining & Knowledge Discovery:
    • principles of the knowledge discovery process;
    • data clustering; * frequent pattern mining;
    • association mining;
    • prediction and sequence mining;
    • Markovian processes, Bayesian networks and graphical models.
  • HPC for Big Data:
    • principles of Big Data processing and HPC;
    • review of main software frameworks (e.g. Hadoop stack);
    • GPU-based processing (CUDA, OpenCL).

Type of exam:  Assessment of individual lab assignment plus overview test.

The courses are taught as a combination of lectures with practical lab components.

Source: Sentinel-1

Learning Outcomes: Upon completion of the module, students are able to:

  • understand the principles of active and multitemporal remote sensing;
  • remember of opportunities offered those recent sensors available in remote sensing;
  • process the data provided by such sensors;
  • perform data analysis to address specific methodological tasks;
  • use dedicated software.

Module content:

  • Lidar:
    • principles (including Multi-Echo, Full Wave Form, Multispectral) and sensors (ALS, MLS, TLS);
    • data processing (DEM and 3D points clouds);
    • use of dedicated software (e.g. CloudCompare).
  • SAR:
    • principles (including Polarimetry, Interferometry) and sensors (Sentinel-1, TerraSAR-X);
    • data processing (speckle reduction, target detection, land cover mapping);
    • use of dedicated software (e.g. SNAP).
  • Time Series & Video:
    • principles and Sensors (Sentinel-2, video from UAV);
    • data processing (change detection, object tracking, land cover mapping);
    • use of dedicated software.

Type of exam:  Assessment of individual lab assignment plus overview test.

The courses are taught as a combination of lectures with practical lab components.

Learning Outcomes: Upon completion of the module, students are able to:

  • understand main concepts behind human-computer interaction;
  • design effective GUI;
  • elaborate visualization strategies to ease understanding of the data.

Module content:

  • Interaction:
    • concepts, theories and models of HCI;
    • user experience and GUI ergonomics;
    • GUI assessment and graphical design.
  • Data Visualization:
    • principles and methods;
    • data transformation and dynamic querying;
    • statistical graphics;
    • programming visualization methods;
    • links with GIS.

Type of exam:  Assessment of individual lab assignment plus overview test.

The courses are taught as a combination of lectures with practical lab components.

latest update: December 11, 2018