David J. Harris

icon location Machine Learning Engineer | Data Scientist | 6+ years at Wayfair & Chewy | Specialized in using statistics and ML to solve business problems.

Results 42 issues of David J. Harris

Add the option for Nesterov momentum

2 comment

enhancement
eventually

test prediction/evaluation code

Currently don't actually test that it _works_, just that it doesn't overwrite the original object

test

MRF nonlinearity should optionally report its distance from `tol`

1 comment

enhancement

Building phylogenetic covariance matrices

1 comment

I think this is all I need to do: ``` # Covariance matrix under an OU model base.covar = vcv(rescale(tre2, sigsq = 1, alpha = 0.1, model = "OU")) #...

enhancement

networks with no predictors

1 comment

no predictors, just stochastic nodes in the first layer. Will allow for unsupervised learning about species composition.

enhancement
test
eventually

simple partial dependence plots

enhancement

Convenience functions for extracting network properties

- output - loss - ?

API

Condition predictions on an observed output

i.e. weigh a bunch of predictions by their likelihood of generating what's been observed so far

enhancement
API

Multinomial output

enhancement

Update docuementation for mistnet, network, and layer

These are the key files

documentation