Calvin McCarter

AI Researcher and Engineer

Software

The kernel density integral (KDI) transformation is a feature preprocessing method. It interpolates between the two leading feature preprocessing methods as limiting cases: linear min-max scaling and quantile transformation, offering protection from the weaknesses of each.

GitHub repository
PyPI (pip) package

ConDo performs domain adaptation (also known as batch correction) under confounding. Domain adaptation methods typically transform domains to match each other, but what if there are true differences between the domains that must be preserved? ConDo conditions on confounding variables while finding an optimal transformation.

GitHub repository
PyPI (pip) package

PerturbNet learns the gene network that modulates the influence of SNPs on phenotypes, using SNPs as naturally occurring perturbation of a biological system. PerturbNet uses a probabilistic graphical model to directly model both the cascade of perturbation from SNPs to the gene network to the phenotype network and the network at each layer of molecular and clinical phenotypes. PerturbNet learns the entire model by solving a single optimization problem with an extremely fast algorithm that can analyze human genome-wide data within a few hours.

Zipped package with CLI, Matlab, and Python interfaces
GitHub page for CLI, Matlab, and Python

Open Source Contributions

onnx2pytorch
A library that converts ONNX models to PyTorch.
[main contributor]

MLPerf Inference
MLPerf™ Inference Benchmark Suite.
[memory-efficient implementation of pyramidal encoder in RNN-T model]

matrix-completion
Python library for in-memory classical matrix completion.
[implementation of memory-efficient incremental SV thresholding]

EnsTop
An ensemble approach to topic modeling using pLSA.
[implementation of sample-weighted pLSA]

PyTorch
The library that freed us from TensorFlow.
[added lazy InstanceNorm]

tf2onnx
A library that converts TensorFlow models to ONNX.
[statically-determined output padding for ConvTranspose]

nanopq
Python library for Product Quantization (PQ) and Optimized Product Quantization (OPQ).
[OPQ initialization using multivariate Gaussian assumption]