-
Heap in R and python
Empirical time complexity
-
Ward clustering
Review of existing segmentation and clustering
-
Agglomerative hierarhical binary segmentation
Clustering using loss or distance minimization
-
Speed of named versus numbered indexing in R
Rotated Hi-C data visualization
-
Where does subtrain come from?
Review of references on cross-validation
-
Stratified batch sampler for mlr3torch
Demonstrations and verifications of correctness using sonar data
-
Paris tourist suggestions
Fun spots
-
MPI for parallelization in R
Another method for machine learning experiments
-
A custom DataLoader for mlr3torch
Stratified sampling for imbalanced classification
-
Creating large imbalanced data benchmarks
Tutorial with OpenML
-
Load-balanced parallel machine learning benchmarks
A new approach using filelock and batchtools
-
Debrief May 2025
Zurich, Munich, Mons
-
Parallel machine learning benchmarks
A new approach using targets and crew.cluster
-
New parallel computing frameworks
batchtools, clustermq, rush, mirai, crew, targets
-
Centralized vs de-centralized parallelization
Exploring rush
-
mlr3 tutorials
Links to other blogs
-
Comparing change-point pruning methods using square loss
Pruned Exact Linear Time (PELT) and Functional Pruning Optimal Partitioning (FPOP)
-
Organizing computational research projects
Guide for research students
-
Volcano plots
Redesign of SOAK paper results
-
Torch learning with binary classification
Implementing AUM loss in mlr3torch
-
Creating imbalanced data benchmarks
Tutorial with MNIST
-
Comparing neural network architectures using mlr3torch
Convolutional network versus linear model
-
Comparing pruning methods for optimal partitioning
Pruned Exact Linear Time (PELT) and Functional Pruning Optimal Partitioning (FPOP)
-
Are pipe operations linear or quadratic?
A demonstration of atime on mlr3torch
-
Bike ride map and time series data viz
A demonstration of animint2 and sf
-
Configuring eduroam
On cell phones and linux
-
Benchmarking data.table with polars, duckdb, and pandas
Demonstrating advantages of data.table
-
Implementing optimal partitioning in R
Comparison with Pruned Exact Linear Time (PELT)
-
Cross-validation experiments with torch learners
Demonstration of mlr3torch + mlr3resampling
-
Overhead of auto-grad in torch
Comparison with explicit gradients
-
AUC and AUM in torch
Demonstration of auto-grad
-
Ordinary least squares algorithms
Comparing computation time in R
-
Collaborations not allowed
Parsing a web page with regex
-
Code of conduct / conduite
Lecture obligatoire pour participants du labo LASSO / Required reading for LASSO lab participants
-
Visualizing prediction error
And clearly showing differences between algorithms
-
History of supervised change-point detection
Using git bisect to find a survival bug
-
Generate publications page
Parsing bibtex and generating markdown
-
Writing comprehensible tests
Documenting key code magic numbers in animint2 tests
-
Rust versus Go
Similarities and Differences
-
How reproducible are benchmarks?
Comparing atime results on different computers
-
Collapse reshape benchmark
Comparison with data.table
-
Porting base R regex code to nc
Case study with a complex regex
-
Benchmarking a change in data.table
Progress reporting for group by operations
-
Mammouth tutorial
Cluster computing for students at UdeS
-
Research student application
Please read if you want to do research under my supervision
-
HTML to Markdown
Regex for porting my lab web site
-
Short bio
Some text to use for talk introductions
-
Directions to my office in Sherbrooke
With a map in English!
-
New code for various kinds of cross-validation
Cross-validation in R with mlr3
-
Capturing regular expressions
Extracting data from loosely structured text
-
The importance of hyper-parameter tuning
And parallellizing machine learning experiments in R
-
When is it useful to train with combined subsets?
An exploration using cross-validation
-
Parsing check logs using regular expressions
A demonstration of nc R package
-
Unable to load shared object, Undefined symbol
Creating and explaining a linker error
-
Reshape performance comparison
Demonstration of asymptotic timing comparisons
-
Cross-validation with variable size train sets
Determining how many samples are necessary for optimal prediction
-
Upgrading R arrow
More build debugging
-
Partial matching on data frame row names
Comparing efficiency using atime
-
Interpretable learning algorithms with built-in feature selection
Regularized linear model and decision tree
-
Generalization to new subsets in R
Coding non-standard cross-validation
-
Comparing machine learning frameworks in R
for loop, mlr3, tidymodels
-
data.table CRAN diffs
Verifying consistency between CRAN and github
-
data.table asymptotic timings
Motivational figures
-
Debugging python code in emacs
Fixing a bug and building old emacs
-
Count unique students
Regex and data table summarization
-
Essential emacs key commands
Cheat sheet for my students
-
Splitting an R package
Recommendations from experience with spatstat
-
Installing Rmpi on the cluster
This package needs special treatment on compute nodes
-
Segfault using R arrow
Reproducing and fixing an error
-
Re-building vignettes on windows
Fixing mysterious error
-
Modifying default gcc compilation flags
When compiling R packages
-
Installing Ubuntu on an old Mac
Step by step instructions
-
spack package manager
contrast with conda
-
Checking R package on M1 Mac
Web services for R package developers
-
Comparing asymptotic timings of CSV read/write functions
Some surprising differences
-
Debugging C code
valgrind and gdb are essential tools
-
CRAN Meta-data
Backing up MRAN
-
Cross-validation experiments on the cluster
NAU monsoon tutorial
-
Generalization to new subsets
Cross-validation in python
-
R Package Release History
Extracting and plotting data from CRAN web site
-
Submitting python jobs on monsoon
And anaconda setup
-
Cloud Storage
Different options for internal and external sharing
-
Indirect reverse dependencies
Computing the entire graph, and histogram tutorial
-
GUI for WSL on Windows 10
use cygwin instead of vcxsrv
-
Reformatting NEWS files
Regular expression example
-
R packages on github
How to query CRAN meta-data
-
Positive and negative log transform
Non-linear transformations for heat maps and signed p-values
-
Research Mentorship Plan
Required reading for potential students
-
Historical reverse imports
Analysis of R package usage over time
-
Learning with Area Under the Min
How to use torch with a non-standard loss
-
Torch randomness
Reproducible neural network learning
-
No argument unpacking in C
But there is in R and Python
-
AUM in Torch
Auto-grad of a non-differentiable loss function
-
Erdos number
A distance calculator
-
Plotting the probability simplex
An application of matrix inversion
-
Link-time optimization
Fixing warnings from CRAN checks
-
Finding symbols in object files
Using objdump to find cerr
-
Ten years of R project in Google Summer of Code
Some success stories from my participation
-
Simple methods for defining small data by row
Comparison with base R and tribble
-
The C book
Documentation of stringize macros
-
Evidential machine learning
An alternative to probability
-
Stress testing reshape operations on list columns
Advantages of updated data.table::melt
-
Defining data by row and regex by sub-pattern
Avoiding separation of related concepts in code
-
Update about data reshaping and visualization in R and python
data.table, tidyr, nc, pandas, datatable, plotnine, altair, bokeh
-
Convex clustering theory
Recent results on trees and cluster shapes
-
R packages that depend on system libraries
How to pass CRAN checks
-
The UCR Time Series Archive
A benchmark for classification algorithms
-
Multi-threaded sorting
Thread safety of qsort variants
-
Faster AUM computation?
Log-linear C++ STL containers vs linear time radix sort
-
New ideas for classification
Weston-Watkins multiclass SVM and AUC optimization
-
Emulating the python interactive console
My hack using the code module
-
New packages for data storage and reshaping
tidyfast, tidyfst, fst, arrow, feather, parquet
-
Computing K-means train/validation error
Alternatives to for loops in R
-
Parsing CRAN maintainers
Regular expressions using nc R package
-
emacspeak
Teaching my son to type in emacs
-
Random train/validation/test assignment
Different methods tried by my students
-
C/C++ completion in emacs
Configuration details
-
Data manipulation libraries
Translating between data.table, pandas, dplyr
-
Custom evaluation metrics in TensorFlow
Implementing the exact area under the ROC curve
-
Fast parameter exploration
Caching and parallel execution
-
binsegRcpp inside a C++ program
Embedding Rcpp code into a main function
-
Embedding R
Compiling a program that links to R
-
R batchtools on Monsoon
Cluster computing tutorial for NAU students
-
Arizona time
Why does internet tell people the wrong time?
-
Emacs local variables
Custom configurations for R
-
Ubuntu setup and LaTeX debugging
Installing and configuring a 10 year old Mac
-
X forwarding on windows
Installing and configuring cygwin
-
Scientific poster suggestions
A helpful video
-
Tinyverse
Complex software dependencies considered harmful
-
useR 2019 debrief
Interesting talks I saw in Toulouse
-
R in Docker on Mac
Reproducing valgrind messages using an R-hub image
-
R package installation on windows considered harmful
Warning for unsuccessful DLL copy should be an error
-
tikzDevice on windows
Fixing missing packages
-
future.batchtools
Simple parallel R code on a computer cluster
-
OpenMP
Simple parallel for loops in C++
-
gdb with R
how to find line numbers of assertion errors
-
Eigen and UNDEBUG
Turning on runtime assertion errors for compiled code in R packages
-
survivalsvm
Support vector machine for survival analysis
-
Testing PeakSegPipeline on Travis with SLURM
Also batchtools and texinfo
-
Tweet when donation received
My first google script
-
Setting default web browser in LXDE
Need to create a .desktop file
-
Keyboard remapping on windows
Changing caps lock to control on windows
-
Training Benchmark for Deep neural networks
A new benchmark data set for neural network training
-
The biglasso package
An on-disk implementation for huge data
-
True reproducibility in R
The switchr package and manifests
-
Seasonal temperature variations where I have lived
Using R to download and plot temperature data from wikipedia
-
Loon
-
What Science Is
-
Compiling R
-
R-GSOC-2017
R project in Google Summer of Code 2017
-
useR! 2017 debrief
summary of interesting work I saw in Brussels
-
Combining data tables in R
rbind inside the for loop is much slower than outside
-
new web site
now more complete and informative
Newer
Older