/EXCEL-for-Deep-Learning

Primary LanguageJupyter NotebookMIT LicenseMIT

Personally, I find it is very useful to use EXCEL to visualize abstract ideas. Sometimes, complex maths notations can be expressed in simple EXCEL functions.

Note: For gradient decent, please ensure to have 'Solver' as add-on and choose "GRG Nonlinear" as solving method.

Visualized Dropout (value between 0.1 to 0.9) for CNN

  • Dropout.xlxs

Visualized gamma values in relation to losses

Reference: Focal Loss for Dense Object Detection http://arxiv.org/abs/1708.02002

  • Focal_loss_chart.xlxs
  • Focal_Loss.ipynb

Experiment gradient explosion

  • CycleGAN.xlxs

Visualization for WGAN (comparing L1 & L2 losses)

Reference: Wasserstein GAN http://arxiv.org/abs/1701.07875

  • WGAN.xlxs

Visualization of up-sampling using different methods and kernel sizes

  • Up_sampling.xlxs