No materials for the event yet, sorry!

In 2019 I started developing LearningML, an educational platform intended to facilitate the teaching of Machine Learning fundamentals through hands-on activities. With this tool, students can generate ML models for text, image and numerical set recognition and make software applications that incorporate these models. The platform consists of an ML model editor, where the student performs the 3 main phases of supervised ML: 1 - collects example data (training), 2 - runs an algorithm to build the ML model (learning) and 3 - evaluates the resulting model (evaluation). The ML editor allows to build a model without programming. It is therefore very suitable, as demonstrated in [1], for students from 10 to 16 years old. Afterwards, and optionally, the student can code an application that incorporates the ML model thanks to some ML blocks added to a modified Scratch instance.

As a continuation of the project, and with the same objective of teaching the basics of Machine Learning, but focusing on coding, I have developed lml-Snap! a modification of Snap! that incorporates new blocks designed to carry out all the phases of supervised ML from the coding editor itself. In other words, both the construction of text, image and numerical set recognition models, and the programming of applications that incorporate these projects, are carried out programmatically. In short, the ML model editor is dispensed with and the student is left to program his or her own model building strategy. This approach favours computer programming practice, especially the use and manipulation of lists and file handling, as training data often comes from external structured files (JSON, CSV) that are loaded into variables and converted into lists.

At the moment the capabilities of lml-Snap! and LearningML are identical. In both cases the ML algorithms, although configurable, are predefined. The only difference is that the construction of the ML model is done programmatically in lml-Snap! However, the next step in the development of lml-Snap! provides the possibility for the student to build their own ML algorithms, further extending the expressiveness and control over the whole supervised ML process.

  1. Rodríguez-García, J. D., Moreno-León, J., Román-González, M., & Robles, G. (2021, March). Evaluation of an online intervention to teach artificial intelligence with learningml to 10-16-year-old students. In Proceedings of the 52nd ACM technical symposium on computer science education (pp. 177-183).
20 min
Auditorium (Online)
Snap!Con 2023
Presented via:
In person