All these things will be lost like tears in rain.
1 / 68
Sep 3, 18
Sep 3, 18
Feb 9, 17

Training your own MNASNET

Or how we failed

Speeding up word distance calculation 100x

My plain approach to faster and more practical cosine distance calculation

Solving class imbalance on Google open images

Or potentially building a model cascade to alleviate severe class imbalance

Playing with Crowd-AI mapping challenge - or how to improve your CNN performance with self-supervised techniques

A small case for searching for internal structure within the data, weighting and training your CNNs properly

Playing with Variational Auto Encoders - PCA vs. UMAP vs. VAE on FMNIST / MNIST

TLDR - they are very cool - but useful only on very simple domains and datasets

Playing with adversarial attacks on Machines Can See 2018 competition

Or how I ended up in a team that won in the Machines Can See 2018 adversarial competition

Exploring the limits of unsupervised Machine Learning in Computer Vision

Or continuation of flat saga (there is no free lunch indeed)

Applying Deep Watershed Transform to Kaggle Data Science Bowl 2018 (dockerized solution)

And why this competition was a lottery

Playing with a simple SOCKS5 proxy server on Digital Ocean and Ubuntu 16

A step by step explanation for non-technical users

Playing with electricity - forecasting 5000 time series

Applying random forests and deep encoder-decoder RNNs to time series prediction