Random Variables

Quotes

“Data is not information, information is not knowledge, knowledge is not understanding, understanding is not wisdom.” – Clifford Stoll

“A human being should be able to change a diaper, plan an invasion, butcher a hog, conn a ship, design a building, write a sonnet, balance accounts, build a wall, set a bone, comfort the dying, take orders, give orders, cooperate, act alone, solve equations, analyze a new problem, pitch manure, program a computer, cook a tasty meal, fight efficiently, die gallantly. Specialization is for insects.” – Robert Heinlen (thanks RT)

“If I had an hour to solve a problem, I’d spend 55 minutes thinking about the problem and 5 minutes thinking about solutions.” – Albert Einstein

“Simplicity is achieved by resisting the urge to make things complex.” – James Gosling

“Le mieux est l’ennemi du bien (The better is enemy of the good).” – Voltaire

“Prediction is very difficult, especially about the future.” – Neils Bohr

“Many of today’s analytical insights are nothing but water-cooler trivia.” – Sean P. McKenna

Geeky Puns

“Hey, what’s your name?”                 “It’s Ferris [sic], but ever since I spent that weekend in the rain, people just call me Rusty” (an SPM original)

Rage Against the Machine Learning! (an SPM original)

Super PAC learning (courtesy of KPD)

Observations and Quips

Full Motion Video – A rose by any other name …

What’s with the hype around “Full Motion Video” (FMV) (for example, here)?  Merriam-Webster defines FMV as “the display of video images at a rate (such as thirty frames per second) at which objects appear to move smoothly and continuously.  So, it’s what we have always known as “video,” just with a fancy name.

“Weight Sharing” in convolutional neural nets – the low-down

I’ve been working with convolutional neural nets (CNNs) lately and I’ve encountered this phrase “weight sharing” in various places.  To me, this sounded like the kernel used for the convolutions, a.k.a. the “weights” underwent some operation that amounted to “sharing.”  Maybe part of the image got this, another part got that, … you know, that sort of thing.  Nope.  All “weight sharing” means is there is one set of weights (i.e., a single kernel) that is used across the entire image.  Uh.  So, why not call it something like “fixed weight” or “constant weight?!”