Ryan Burn @RyanBurnML

#math #engineer | exploring how to apply optimization algorithms to build better models: https://t.co/gyO7mL8eTL ryanburn.com San Francisco, CA Joined May 2012

Tweets

142
Followers

52
Following

110
Likes

167

Towards Data Science @TDataScience

a year ago

For a thorough and accessible introduction to objective Bayesian hypothesis testing, we encourage learners to explore @RyanBurnML's explainer, which balances theory and practical considerations. buff.ly/45nTObs

0 2 10 3K 1

Ryan Burn @RyanBurnML

3 years ago

What's the deal with your bookings @Delta? Every time I try to book a flight and apply ecredits the webpage locks up and becomes unresponsive.

0 0 0 0 0

Ryan Burn @RyanBurnML

3 years ago

wtf @Windows? This is not the UX I'm looking for: github.com/microsoft/WSL/…

0 0 0 0 0

Towards Data Science @TDataScience

4 years ago

Logistic Regression and the Missing Prior by @RyanBurnML buff.ly/36aFgS5

0 10 55 0 9

Ryan Burn @RyanBurnML

4 years ago

#machinelearning #statistics buildingblock.ai/logistic-regre…

0 2 1 0 0

Towards Data Science @TDataScience

4 years ago

In his new post, @RyanBurnML explains how to build a Bayesian ridge-regression model and handle the hyperparameter that controls regularization strength. buff.ly/3BMHpyZ

0 3 14 0 9

Ryan Burn @RyanBurnML

4 years ago

Want to use #cpp20 modules with @bazelbuild and clang-15? Here's how 👇 buildingblock.ai/cpp20-modules-…

0 0 0 0 0

Ryan Burn @RyanBurnML

4 years ago

#statistics #machinelearning buildingblock.ai/bayesian-ridge…

0 0 1 0 0

Ryan Burn @RyanBurnML

4 years ago

Reflection is a feature #cpp really needs. Unfortunately, a proper language solution is still years away. In the meantime, here's how you can reflect with an inline macro buildingblock.ai/reflection

0 0 0 0 0

ITNEXT @ITNEXT_io

4 years ago

Why Standard C++ Math Functions Are Slow by @RyanBurnML at #ITNEXT. #Cplusplus #Performance itnext.io/why-standard-c…

0 2 2 0 1

Ryan Burn @RyanBurnML

4 years ago

Forming new allocator-aware types by composing existing AA types can lead to a lot of boilerplate code. Using reflection, here's how you can do it with minimal maintenance burden #cplusplus 👇 buildingblock.ai/allocator-awar…

0 0 1 0 0

Ryan Burn @RyanBurnML

4 years ago

👏Multidimensional #cpp23 array support being added to gcc: gcc.gnu.org/pipermail/gcc-…

0 0 0 0 0

Ryan Burn @RyanBurnML

4 years ago

Leave-one-out cross-validation is accurate but expensive. For logistic regression and other GLMs, approximate leave-one-out cross-validation is a much less expensive metric that's nearly as good. This brief guide explains how to use it. #statistics #machinelearning

0 0 0 0 0

Download Image

HackerNoon | Learn Any Technology @hackernoon

4 years ago

"Leave-one-out Cross-validation (LOOCV) is one of the most accurate ways to estimate how well a model will perform on out-of-sample data", @RyanBurnML. hackernoon.com/how-to-use-app…

0 1 3 0 0

Download Image

Ryan Burn @RyanBurnML

4 years ago

I find it disappointing #cpp20 concepts don't allow you to write code like this. The compiler should have everything it needs to know to instantiate the myptr class.

0 0 0 0 0

Download Image

Ryan Burn @RyanBurnML

4 years ago

This bulk rename feature from fern is really improving my #vim workflow. 👇 github.com/lambdalisue/fe…

0 0 0 0 0

Ryan Burn @RyanBurnML

4 years ago

I really hope P2128 makes it into #cpp23 👉 open-std.org/jtc1/sc22/wg21… so that we can write array[i, j, k] obscure uses of the comma operator shouldn't get in the way of sensible syntax

0 0 0 0 0

Ryan Burn @RyanBurnML

4 years ago

Put together a logistic regression example that shows how to 1. use second-order optimization to quickly find the hyperparameters that optimize approximate leave-one-out cross-validation 2. use ALOCV values to identify outliers #machinelearning 👇 buildingblock.ai/multinomial-lo…