Impending Algorithmic Revolution
sciences of the formula
sciences of the algorithm
new sciences
examples
genomics
neuroscience
economics
networks
data has exploded
data
low entropy
noisy
big
high-dimentional
self improving algorithms
input space -> computer
worst case
average case (uniform)
arbitrary, unknown random source
study as little as possible to past the test
perscriptive machine learning
phases
training phase
input space, typical, and atypical inputs
when typical, find approximate nearest neighbor
self improving sorter
limiting complexity
storage
self improving clustering
2-median over hamming cube
minimize total distance to center point
online data reconstruction
error correcting codes
put a filter before computer
only works with convex
make the data convex!
fix it to make it true
change it as little as possible to make it true
filter between data and program
example
filter has to return a human
returns g(x) because its a human
if f(x) is not a human, must change it a small bit
high dimensional data
dimension reduction
minimize distance difference in projection
math for math
distributed algorithms
flock simulation
each algorithm acts on its own without internal knowledge of others
If Google is a religion, what is its God?
|
| |