This page was created to share some of the things I've learned while studying Machine Learning. Most of the topics here were discussed at LEARN (USP's League of Studies about Machine Learning) or are part of my research on data augmentation for Natural Language to SQL translation models.
Linear Regression is the first model I learned. In this post, I share the main ideas behind linear regression libraries as well as an implementation from scratch.
Read MoreNeural Networks utilize simple building blocks to create powerful models for classification and regression problems. The goal of this post is to introduce Perceptrons and their components to prepare for more complex Neural Network models.
Read MoreK Nearest Neighbors or simply KNN is a family of classifier algorithms that is conceptually easy to understand and to implement and nonetheless may achieves great results. Here is an overview of the algorithm with examples.
Read MoreI have recently started a research project involving Machine Learning to make database queries from sentences in Portuguese. More specifically I'll work on data augmentation to improve models' performances.
Read MoreIn this research update I share what I learned in the first couple of weeks. Mostly, I read articles to familiarize with text-to-sql problems and models and to understand what are the benchmark datasets used for this task.
Read MoreData augmentation for text involves a diverse range of techniques as well as some challenges. This post explores simple implementations of back-translation, augmentation with LLM and synonym substitution.
Read More