Topic outline

  • The lecture will be conducted remotely via Zoom:

    https://aalto.zoom.us/j/62964718561
     
    Meeting ID: 629 6471 8561 

    TA: Hossein Firooz (hossein.firooz@aalto.fi)

     
    Modules

    This lecture promotes understanding of good practices for machine learning with noisy and inaccurate data.

    In particular, we focus on feature extraction/ feature subset selection, handling high dimensional data, ANN + Deep Learning, Probabilistic graphical models, Topic models; as well as Unsupervised learning and clustering, Anomaly detection and Recommender systems.

    The lecture exercises are project-based and foster hands-on experience with machine learning approaches on relevant data sets.

    The lecture schedule is given below:

    schedule


    The course is organized in a modular design. Students are free to select any combination of modules. Credits are granted according to the completion of modules. Grading is possible with some modules.

    • Lectures (1 cr): Physical attendance and active participation in at least 10 contact teaching sessions
    • Video lectures (0 cr): Video lectures are provided for some topics to further the learning of the topic.
    • Tutorials (1cr, graded): Preparation, training and presentation in front of the class of an in-depth tutorial on a specialized theme
    • Projects (2cr, graded): Guided Project work in teams of 3 students on a research or industry-relevant topic/data. Video-status reports throughout the project duration and final presentation in form of a poster.
    • Report (2cr, graded, precondition: project): Written technical report on the project work and outcomes. In case of novel results, submission to a scientific venue can be supported.
    • Oral exam (1cr, graded): Individual technical discussion (20 minutes) on the topics taught in the class
    • Practical training sessions (0cr): Contact session and online resources to practice common ML tools
    • Supervision (1cr): Technical supervision of BSc student project group

    It is recommended to complete the ELEC-E7260 course with modules worth 5 credits. However, any combination of modules and credits is possible. In such case, module description (and code) can be adapted to accurately reflect the studied content. Based on the success in the respective modules, students score points which then translate into the final grade:
    Grading scheme in 2021
    The lecture will feature invited lectures given by international experts from industry and academia.