I was reading a book on the life of the italian Jesuit priest Matteo Ricci in China (1582–1610). During that period, he taught the chinese about the Art of Memory used in Europe and contributed some of the images from his memory palace for a book about the teachings of christianity in chinese.
A curious fact is the time that was necessary for a letter from China to arrive in Europe by means of the Portuguese and Spanish trade routes—seven years. Given that my e-mails can arrive practically instantly, I was baffled at the thought of waiting that long to get a response.
The Jesuits in China knew enough of the sea’s dangers to send each of their letters to Europe in two copies—one via Mexico on the Spanish galleons out of Manila, and one via Goa on the Portuguese carracks leaving Macao. Ricci’s superior Valignano may have been startled that one of his letters to Rome took seventeen years in transit from Macao, but Ricci accepted six to seven years as the norm for receiving an answer to a given letter.
This paper talks about how most introductory physics textbooks are awful and provide unrealistic approximations, with no comments on their applicability.
The observant student knows—from exploring the world—that golf drives rise quickly and almost straight and drop parabolically only near the end. This student learns that the world described by her physics textbook is not the real world and that careful observation is irrelevant to physics.
Contrast the experience of the curious student with that of a student who parrots equations and regurgitates textbook paragraphs. This student is untroubled by the golf problem or its variants listed in Table I because he knows the easily memorized “fact” that all trajectories are parabolae. […]
If you ever took a physics class, I’m quite sure you recognize that teaching style. The following figure shows the trajectory of a golf ball considering air resistance, in case you’re interested:
Following my first post on books between the first quadrimester (Jan-Apr), here’s the list for May-Aug. There’s also a post on the books I read read during 2014.
I didn’t read as much as I wanted… but being able to measure how much I’m reading is nice, and I’m coming up with ways to improve on that, e.g. don’t read a book on Haskell and a treatise on Memory written by a historian in parallel.
(I’ve recently discovered that quadrimester is sometimes used as a synonym for quarter in english. Well. In 2016 I might try breaking it into Q1, Q2, Q3, and Q4.)
tl;dr: building features manually is inefficient and extracting them automatically is possible and, in a sense, better.
I’m not writing a lot here these days, so let me summarize: I left my job recently to start working on my Masters at Universidade of São Paulo (USP). I’m studying an area of Machine Learning called Deep Learning (or Representation Learning, the term I prefer) for my research. This post is an overview of what I’ve read so far about the subject.
(By the way, it is practically impossible to enter a PhD program in Brazil without a MSc.)