Welcome to Constant Etherpad!
These pads are archived each night (around 4AM CET) @
http://etherdump.constantvzw.org/
An RSS feed from the etherdump also appears on
http://constantvzw.org/
To prevent your public pad from appearing in the archive and RSS feed, put or just leave the following (including the surrounding double underscores) anywhere in the text of your pad:
__NOPUBLISH__
Changes will be reflected after the next update at 4AM.
overfitting -
theta -
chain rule -
softmax - statistical function
backpropagation -
neuron -
dimensions - number of features in a vector
vector - list of features, mostly expressed in a numerical value
matrix - a set of vectors, with as many rows and columns as dimensions (d) in the vectors (d x d)
skip gram -
glove - specific word embedding software developed by the Stanford University (comparable to word2vec(?)). Richard Sochner is one of the 3 main developers of it.
https://nlp.stanford.edu/projects/glove/
Concepts to explain during workshops
, a possible order of explaning
- different type of problems: classification / regression (information retrieval)
- neural network layers principle
- perceptron/softmax as a simple classifier
- data : where to get, how to get, how to prepare
, how much
- turning words into a matrix / one-hot-vector and other techniques