Open ajschumacher opened 3 years ago
Heckman correction
the kernel trick (formerly #221)
wavelets (formerly #210)
"... much more modern ideas, like the theory of wavelets developed in the 1970s and '80s by Jean Morlet, Stéphane Mallat, Yves Meyer, Ingrid Daubechies, and others" (page 330, Ellenberg)
Turbocodes (formerly #200)
support vector machines (SVM) (formerly #196)
Baum–Welch algorithm (formerly #187)
https://en.wikipedia.org/wiki/Baum%E2%80%93Welch_algorithm
(recommended via that book about Simons)
Fast Fourier transform (FFT) (formerly #154)
Downey's stuff might be a good start... Anyway I saw this tweet:
Which links to this notebook:
https://twitter.com/jezzamonn/status/1083498750050553856
mention Fourier in case I search by that...
https://www.jezzamon.com/fourier/index.html
FFT comes up here: https://www.quantamagazine.org/mathematicians-discover-the-perfect-way-to-multiply-20190411/ kind of surprisingly?
https://realpython.com/python-scipy-fft/
This has a FFT section: https://pythonnumericalmethods.berkeley.edu/notebooks/Index.html
twiddle factors?
Hungarian association algorithm (formerly #153)
particle filters (formerly #151)
transducers (formerly #150)
Kalman filters (formerly #149)
Least-squares estimation: from Gauss to Kalman https://www2.pv.infn.it/~fontana/download/lect/Sorenson.pdf
I had some old note like this:
Wiener process / Kalman filtering (connection with Gaussian process?)
Fun demo recommended by Gerardo Leal: https://www.cs.utexas.edu/~teammco/misc/kalman_filter/
IRR function? in Excel? (formerly #131)
beam search / viterbi (formerly #109 and also #81)
top
, htop
, etc. (formerly #101)
softmax temperature (formerly #105)
and let's group this with softmax generally (formerly #74)
inverting / pseudo-inverting a matrix (formerly #82)
fun highlights of various intuitions: https://twitter.com/michael_nielsen/status/1073673971583705089
tSNE/UMAP (formerly #41)
"Relative Weight Analysis"?
This fairly bad post has gotten some play online:
https://towardsdatascience.com/key-driver-analysis-in-python-788beb9b8a7d
It seems to be largely based on this better article:
Kind of neat as another kind of feature importance?
Learn more languages? C++? Elixir?
figure out a good way to do speech-to-text? for calls and so on?
https://en.wikipedia.org/wiki/Toeplitz_matrix for convolution
Power law: scaling to distribution?
Things that maybe I should learn more about and/or write explainers for...