Sitemap

Page not found. Your pixels are in another canvas.

About

This is a page not in th emain menu

Experimenting with a Custom Distance Based on Haversine with K Nearest Neighbors

In this blog post, I will discuss: (1) the Haversine distance, a distance metric designed for measuring distances between places on earth, (2) a customized distance metric I implemented, “HaversineEuclidean”, which I felt would be more appropriate in an analysis of the California Housing data, and (3) how to implement this custom metric in a K Nearest Neighbors Regression (KNN) with scikit-learn. The implementation is also relevant to other functionalities in sklearn.neighbors.

A Guide to Installing TorchDrug 0.1.0 (CPU) and Using with Jupyter Lab

TorchDrug, a platform for drug discovery with PyTorch, was just released. It was developed at MILA with Jian Tang as the principal investigator and launched on GitHub by Zhaocheng Zhu (full team here).

Graph Neural Networks Spatial Exploration

Ryan L. Murphy, Balasubramaniam Srinivasan, Vinayak Rao, Bruno Ribeiro Janossy Pooling: Learning Deep Permutation Invariant Functions for Variable-Size Inputs at ICLR 2019: https://arxiv.org/abs/1811.01900

Ryan L. Murphy, Balasubramaniam Srinivasan, Vinayak Rao, Bruno Ribeiro Relational Pooling for Graph Representations at ICML 2019: https://arxiv.org/abs/1903.02541

Poster presentation on Janossy Pooling at ICLR.

Oral presentation on Relational Pooling to a community of Machine Learning-focused scholars at NetSci.

Oral presentation on Relational Pooling at ICML. Slides available at ICML.