Skip to content

Instantly share code, notes, and snippets.

Keybase proof

I hereby claim:

  • I am winding-lines on github.
  • I am mseritan (https://keybase.io/mseritan) on keybase.
  • I have a public key whose fingerprint is D4D7 D81F E4BC C26E CAF0 9047 2FD8 2F1E A0DA 81DE

To claim this, I am signing this object:

@winding-lines
winding-lines / Cargo.toml
Last active September 27, 2016 16:18
Linking against hyper on both OS X and Linux
[package]
name = "http"
version = "0.1.0"
authors = ["Marius Seritan <github@seritan.net>"]
[target.'cfg(target_os = "macos")'.dependencies]
hyper = {version="*", default_features=false, features=["security-framework"]}
[target.'cfg(target_os = "linux")'.dependencies]
hyper = {version="*"}
function generate_output(pomdp, belief_updater)
@printf("Entering\n")
Random.seed!(2)
# reset the policy
p = ToEnd(0) # here, the argument sets the time-steps elapsed to 0
# run the simulation
c = CairoRGBSurface(700, 500);
@winding-lines
winding-lines / RL-acronyms.MD
Last active November 24, 2018 14:34
RL Acronyms

arxiv POMCPOW - partially observable Monte Carlo planning with observation widening PFT-DPW - particle filter trees with double progressive widening

cmu PBVI - point based value iteration

rhoPOMDP - POMDP with belief dependent rewards nips

@winding-lines
winding-lines / Jupyter Doctor.MD
Last active November 24, 2018 16:59
Various Jupyter related scripts

Basic diagnostics

Start New Python notebook and try running 1+1. Then start Julia notebook with the same.

If the javascript console has errors around downloading js files you can diagnose WebIO problems - from stackoverflow

import os.path
from jupyter_core.paths import jupyter_path
@winding-lines
winding-lines / Eval.md
Created July 16, 2019 14:25
Stats, ML

The Relationship Between Precision-Recall and ROC Curves

PR

  • Precision vs Recall
  • better with highly skewed datasets

ROC recommended by Provost et al. (1998)

  • Receiver Operator Characteristic
  • True Positive Rate (fraction of positive examples correctly labeled) vs False Positive Rate (fraction of negative examples misclassified as positive)
@winding-lines
winding-lines / Dockerfile
Created January 5, 2020 17:28
Build python-pcl
# Based on https://github.com/alliecc/argoverse_baselinetracker
ARG CUDA="10.0"
ARG CUDNN="7"
FROM nvidia/cuda:${CUDA}-cudnn${CUDNN}-devel
# install basics
RUN echo 'debconf debconf/frontend select Noninteractive' | debconf-set-selections \
&& apt-get update --fix-missing -y \
&& apt-get install -y apt-utils git curl ca-certificates bzip2 cmake g++ pkg-config \