Learning for eLearning

Abstract: Finding the right solution to a business case goes beyond the discovery of the most accurate algorithm; sometimes it looks like a form of art, a delicate balance between data mining and feature engineering (dataset for training), data science (optimal algorithm and training pattern), computer science (proper delivery of the algorithm), and computer engineering (ideal architecture to deploy the algorithm).

Navigating through this world is not easy, especially for companies that are beginning implement a layer of data science to cater to and enable their business cases.

In this presentation you will be guided in a journey, from an idea, to its application, and its implementation and delivery, passing through bumps and change of directions, with a constant focus of finding the perfect answer to a business problem.

In particular you will learn about the development of a recommendation system for eLearning, in order to suggest formal or informal training contents to users, but also to indicate users potentially interested in a specific content.

You will have a better understanding of how Neural Network and Recurrent Neural Network work, and their strength and their weaknesses. You will learn how they compare with shallow algorithms like ensemble and boosting algorithms. Furthermore you will see example of preprocessing and feature engineering to adapt your data to the chosen model.

A second business case that will be presented will be the automatic generation of video content from audio recording. Important steps in the pipeline will be extraction of key terms and images related, video editing, and subtitles, for which you will hear a few ideas and suggestions.

At the end of the presentation you will be able to consider your business case as a whole, and hopefully find the most successful compromise between feasibility and accuracy.

Bio: Lucia‚Äôs true passion is discovering the right key to crack a new challenge, finding the right hypothesis and unravelling hidden patterns from data, and making the most accurate predictions. Data science was the natural continuation of her post-doc in Neuroscience.

After embracing this new adventure in industry she worked in different fields, from social media analysis, to Law, and finally eLearning.

In all her projects she holds that sparkle, which leads her to learn order from chaos, and deliver solutions to the most diverse problems.