Thursday, June 16, 2016

It's all just math!

An interesting article in Wired that makes data science (deep learning in particular) less mysterious to the outside world.  The more you review how certain modeling frameworks work, the more they become the mundane.  It takes processing power because computers don't make decisions like humans do, but instead take a very statistical and formula-based approach.

The article lists out the fundamental basics of data science as well:

In a sense, deep learning is not unique. Any machine learning system—deep or not—consists of the following fundamental components: Performance element: the component of the system that takes some action in the world (e.g., making moves in the game of Go).
  1. Target function: the function being learned (e.g., a mapping from board positions to move choices in Go).
  1. Training data: the set of labeled data points used to approximate the target function (e.g., a set of Go board positions, each labeled with the move chosen by a human expert in that position).
  1. Data representation: each data point is typically represented as a vector of pre-determined variables (e.g., the position of a piece on the Go board).
  1. Learning algorithm: the algorithm that computes an approximation of the target function based on the training data.
  1. Hypothesis space: the space of possible functions the learning algorithm can consider.