The Curious Engineer

All blogs / AI / ML

Brilliantly wrong

arogozhnikov.github.io

Brilliantly Wrong — Alex Rogozhnikov's blog about math, machine learning, programming, physics and biology.

AI / MLMathSciencePython

’15’20’25
Posts
59 from 20 Jul 2013 to 15 Jul 2026
Posting
now and then (2 in the last year)
Kind
personal blog
Feed
https://arogozhnikov.github.io/feed.xml

2026 2

  1. Beyond Affinity of Proteins
  2. State of Wall in Protein Language Models in 2026

2023 4

  1. Fastest Autograd in the West
  2. Optical pooled screens of cells (overview of emerging biotechnology)
  3. Einops, retrospective of 5 years
  4. Schema migration should be a responsibility of DB

2022 1

  1. Delimiter-first code

2020 3

  1. Things I wish someone told me about microscopy
  2. Don't write command-line interfaces (generate them)
  3. Statistics: numbers that lie to you

2019 1

  1. Twin training: trick for better model comparisons

2018 1

  1. Einops — a new style of deep learning code

2017 3

  1. Clustering applied to showers in the OPERA
  2. Machine learning applied to showers in the OPERA
  3. Machine Learning in Science and Industry

2016 10

  1. Hamiltonian MC explained
  2. Jupyter (IPython) notebooks features
  3. MLHEP 2016 lectures slides
  4. Occam razor vs. machine learning
  5. Gradient boosting playground
  6. Gradient boosting explained
  7. .new_item for python lists
  8. Interactive demonstrations for ML courses
  9. Testing implementations of LibFM
  10. Reconstructing pictures with ML

2015 22

  1. Optimal control of oscillations
  2. Cumulative sum as a test for your programming environment
  3. Rule of statistical error
  4. Reweighting with Boosted Decision Trees
  5. sPlot: a technique to reconstruct components of a mixture
  6. ROC curve explained
  7. part 2
  8. part 1
  9. Speed benchmarks: numpy vs all.
  10. MLHEP 2015 lectures slides
  11. part 2
  12. Multimixture fitting
  13. part 1
  14. Automatic reweighting with gradient boosting
  15. Decision train classifier
  16. Removing margins of PDFs
  17. Thoughts about tensor train
  18. Anomaly detection at CERN experiments
  19. Some links on deep learning
  20. Numpy exercises
  21. Old-style optimization methods for neural networks
  22. An interesting equality from linear algebra

2014 5

  1. So many names
  2. Cousteau
  3. Optimization of vector operations with bit hacks
  4. Minibus stochastic process
  5. Theano (python library)

2013 7

  1. Numbers and Naturalness, part 3
  2. Legendre transformation without Legendre transformation
  3. Soviet computer "Сетунь"
  4. Redundant numeral systems
  5. Numbers and Naturalness, part 2.
  6. Numbers and Naturalness, part 1.
  7. Why and What