Posts on Roverdi's Website

Multilayer Perceptron 3: Beyond XOR

Thu, 11 Jun 2026 00:00:00 -0300

In the previous posts, we built the MLP foundation: we understood its structure and feedforward and how it learns through backpropagation. Now let’s take that knowledge to more realistic problems and open the horizon to what comes next.

Other Ways to Measure Error

Up to now, we’ve used MAE (Mean Absolute Error) to monitor the network. It’s excellent for its simplicity: it just tells us the average distance between prediction and reality. However, monitoring error and training the network don’t have to use the exact same measure. Depending on the problem, other loss functions can generate more useful gradients for learning.

Multilayer Perceptron 2: Learning and Backpropagation

Mon, 08 Jun 2026 00:00:00 -0300

In the previous post, we saw how the MLP organizes neurons into layers and how data flows through the feedforward pass. By the end, the network was producing predictions with fixed weights and we were measuring the error — but the network wasn’t doing anything with it.

Now let’s fix that: how does the network use the error to learn?

Optimization

Training a neural network is, at its core, an optimization problem: the goal is to find the set of weights that produces the smallest possible error between the network’s output and the expected result. Without an optimization algorithm, weights would remain static and the network would be incapable of learning.

Multilayer Perceptron 1: Structure and Feedforward

Thu, 04 Jun 2026 15:00:00 -0300

In the previous post, we explored the Perceptron: we saw that it can learn and make simple decisions, but we also discovered it has a hard limit. Being a linear classifier, it can only solve problems where the data can be separated by a straight line.

The most iconic limitation in the history of neural networks was the logical XOR (Exclusive OR) operator. Since the points of that problem can’t be split by a single line, a lone Perceptron loops forever and never learns the rule.

Perceptron: The Foundation of Deep Learning

Sun, 31 May 2026 22:00:00 -0300

At its core, all modern artificial intelligence is just trying to imitate what you do without thinking.

When you decide whether to cross the street, or whether that email belongs in the spam folder, your brain is processing signals, weighing evidence, and making a simple decision: yes or no.

In artificial neural networks, everything started as an attempt to replicate that behavior in a mathematical model, giving birth to what we know as the Perceptron — still, to this day, one of the most elegant ideas in computing.

Tibia 1.03 (Alpha): Sprites

Fri, 29 May 2026 10:00:00 -0300

Introduction

Tibia is one of the oldest MMORPGs still online and under active development. Created in 1997 by students of a German university — who later founded CipSoft, it eventually became very popular, especially in Brazil.

Some time ago, I ended up stumbling upon a copy of the Alpha client (version 1.03), which, as far as I know, was the first publicly released version. You can download it here.

Debian on the Toshiba Portégé M200

Tue, 26 May 2026 21:00:00 -0300

Have you ever tried to install Linux on a machine that simply refuses to cooperate?

Years ago, I was given a Toshiba Portégé M200. For those who aren’t familiar, it’s a 2004 Tablet PC with a screen that was absurdly ahead of its time (featuring Wacom pen support) and a design that still catches the eye today. But it has one tiny little problem: there is no way to boot from USB. On top of that, it has no internal CD/DVD drive, and booting from an external drive only works with a specific Toshiba model that you can only find on Japanese auction sites.

Jevons Paradox in the AI Age

Sat, 23 May 2026 23:00:00 -0300

Have you ever noticed that the more technology advances and saves resources, the more we seem to consume?

Well, this contradiction isn’t new. Back in 1865, British economist William Jevons noticed something: while steam engines became more efficient and used less coal, total coal consumption didn’t drop — quite the opposite. As energy became cheaper, the industry found lots of new uses for it, causing the demand for coal to skyrocket like never before. This phenomenon became known as the Jevons Paradox (or, more popularly, the rebound effect).

WoW Arduino Fisher

Wed, 25 Oct 2023 10:00:00 -0300

Windows XP on a Dell Optiplex 380

Sat, 14 Oct 2023 20:00:00 -0300

SDL2, VMware, Linux and Relative Mouse Mode

Wed, 12 Jul 2023 15:00:00 -0300

Hello World!

Wed, 12 Jul 2023 13:00:00 -0300