February 15, 2020

[R] When scaling a machine learning task to multiple nodes/workers, can you also increase the batch size * number of workers?

What happens exactly to the batch size and learning rate when you use multiple nodes (systems) instead of one for the ML task at hand? submitted […]
February 16, 2020

[D] Use batch normalization in GANs but not VAEs

Going through this TensorFlow tutorial on VAEs, I found the following sentence: Note, it's common practice to avoid using batch normalization when training VAEs, since the […]
February 16, 2020

[P] imgaug 0.4.0 released, adding many new augmenters (e.g. jigsaw puzzles and cartoon-style filter)

The release adds many new augmenters, among them: Polar coordinate warping Rain effects Cartoon filter RangAugment (supports only images) Jigsaw puzzle effects (supports only images, segmentation […]
February 16, 2020

[D] Current SOTA of NN for tabular data?

Any deep learning methods that beat XGBoost and LightGBM? ​ Currently I found papers such as https://github.com/wOOL/DNDT. ​ But what is the current fully-differentiable SOTA for […]