Posts

Neural ODEs

Deep Learning

Ordinalry Differential Equations or ODEs are equations with a single indipendent varlaible (usually called time \(t\)) and one or more derivatives of functions defined in…

Sandro Cavallari

Linear Regression

Statistics

Linear Algebra

Linear regression, as discussed by Montgomery and Runger [1], stands out as a widely employed and intuitive machine learning model. Its primary objective is to establish a…

Sandro Cavallari

Normalizing Flows

Deep Learning

Normalizing Flows (NF) represent a potent technique that facilitates the learning and sampling from intricate probability distributions [1] [2]. These models, categorized as…

Sandro Cavallari

Finetune Large Models

Deep Learning

Nowadays, it is common practice to develop new machine learning projects starting from a large pre-trained model and fine-tuning it to the task at hand. Sam Altman, at the…

Sandro Cavallari

Efficient and Scalable Machine Learning Pipelines

Deep Learning

Jobs related to machine learning usually require managing massive datasets. A well-established rule of thumb that applies to most machine learning projects is that the…

Sandro Cavallari

Data Structures

Coding

Data scrtucures are efficent memory construct used to sotre and organize data in an efficent manner. Adopting the right data structure and having efficent access to the…

Sandro Cavallari

Evaluation Metrics for Ads Ranking Systems

Statistics

Ads ranking systems are the barebone of many modern business and became one of the main success story of machine learning applied to real-world problems. Given an…

Sandro Cavallari

Variational Autoencoders

Deep Learning

In this article, I will delve into the theoretical foundations of Variational Autoencoders (VAE). You can find the code used for both Convolutional Neural Network (CNN) and…

Sandro Cavallari

Basic Principles of Linear Algebra

Linear Algebra

Linear algebra is the branch of math and statistics that is devoted to the study of matrices and vectors. As such, it is broadly used to model real-world problems in…

Sandro Cavallari

Gradinet Descent and Backpropagation

Deep Learning

Most deep learning algorithm relay on the idea of learning some useful information from the data to solve a specific task. That is, instead of explicitly define every single…

Sandro Cavallari