A blog designed like a scientific paper

Laplace's Demon and the Scientific Method

May 28, 2017

Miguel González-Fierro

In 1814, Laplace described an entity that would be able to calculate any future event if it knew the precise location and speed of every atom in the Universe, based on the principle of cause and effect. This will be known as Laplace's Demon. At that time, scientists believed that they could solve all the mysteries of the Universe. Decades later, with Heisenberg's uncertainty principle and other advances, that dream became to an end.

How to deploy an Image Classification API based on Deep Learning

April 17, 2017

Ilia Karmanov and Miguel González-Fierro

Deep learning is notable for surpassing human level performance in image classification. In this post we will explain how to use a pretrained Convolutional Neural Network to classify images and how to deploy this classifier to the cloud through an API.

The Sorcerer's Apprentice

March 4, 2017

Miguel González-Fierro

The Sorcerer's Apprentice was a tale written by Goethe in 1797, popularized by Disney in the 1940 animated film Fantasia. In it, a young sorcerer wants to seize the power of his old master but realizes that he lacks the knowledge to control it. The story teaches us a lesson about the perils of a student thinking he has knowledge when he doesn't, however, when does an apprentice become a master? In this post we will address that question.

A Gentle Introduction to Text Classification and Sentiment Analysis

Jan. 31, 2017

Miguel González-Fierro

The area of text classification has attracted a lot of interest from both the machine learning research community and the industry. One popular application of text classification is sentiment analysis, whose objective is to guess the positive or negative attitude of a user towards a topic given a sentence. In this post I will give an overview of how to apply machine learning techniques to text classification and sentiment analysis.

Learnings from AI & Machine Learning Conference NIPS 2016

Dec. 22, 2016

Andreas Argyriou, Miguel González-Fierro

NIPS conference has become one of the most important events for machine learning research. With a record participation of more than 6000 attendees this year, the field consolidates its importance in both academia and industry. In this post we will discuss the highlights of this year's NIPS conference.

How to Develop a Data Science Project using the Lean Startup Method

Nov. 27, 2016

Miguel González-Fierro

Lean Startup has been a breakthrough during the last years in the entrepreneurial landscape. But this movement is not only a methodology that can be used to create successful businesses, it can also be applied to other domains like Data Science. In this post, I will discuss how to apply the Lean Startup method to a Data Science project. As an example, I created a project that visualizes all football matches that took place in the UEFA Champions League since 1955.

A Gentle Introduction to Convolutional Neural Networks

Sept. 30, 2016

Miguel González-Fierro

Convolutional Neural Networks (CNN) are one of the key components in the success of Deep Learning and the new Artificial Intelligence revolution. They are specially advantageous in tasks such as object detection, scene understanding and, recently, natural language processing. In this post I will explain what are CNN, what is the intuition behind them and provide an example of its performance in character recognition.

When to use Deep Learning in a Data Science Problem

Aug. 10, 2016

Miguel González-Fierro

Deep Learning has been around since 2006 and has produced some of most exciting advances in Machine Learning. However, it is not widely used in the industry by Data Scientists yet. In this post we will discuss this business opportunity and when Deep Learning can be used in a Data Science problem.

How human intelligence works and why that makes us racists

June 25, 2016

Miguel González-Fierro

Have you ever stopped to think how people can be so stupid? Lately, it looks like the world is going crazy. More and more people are supporting radical views. But this is not new, our history is full of atrocities that can be summarized in one simple reason: hatred of the different. The bad news is that our intelligence is optimized to detect differences. The bigger the difference, the higher the effort our brain makes. And many people don't like efforts.

A Gentle Introduction to the Basics of Machine Learning

June 5, 2016

Miguel González-Fierro

Machine Learning is one of the most important technologies nowadays. It is the key element that allows a computer to learn. It is in fact a subfield of Artificial Intelligence (AI). Right now AI is affecting all industries, from finance to medicine, from aerospace to e-commerce. All technological giants such as Microsoft, Google, Amazon, etc. are investing large amounts of money in the area. This post intends to explain the main concepts of Machine Learning to the general public.

10 portfolio projects to get you an AI job

Sciblog - A blog designed like a scientific paper

Miguel González-Fierro

Ilia Karmanov and Miguel González-Fierro

Deep learning is notable for surpassing human level performance in image classification. In this post we will explain how to use a pretrained Convolutional Neural Network to classify images and how to deploy this classifier to the cloud through an API.

Miguel González-Fierro

Miguel González-Fierro

Andreas Argyriou, Miguel González-Fierro

Miguel González-Fierro

Miguel González-Fierro

Miguel González-Fierro

Deep Learning has been around since 2006 and has produced some of most exciting advances in Machine Learning. However, it is not widely used in the industry by Data Scientists yet. In this post we will discuss this business opportunity and when Deep Learning can be used in a Data Science problem.

Miguel González-Fierro

Miguel González-Fierro